release-year-prediction-on-msd-using-neural-nets.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['year_prediction.csv']\n"
     ]
    }
   ],
   "source": [
    "# This Python 3 environment comes with many helpful analytics libraries installed\n",
    "# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n",
    "# For example, here's several helpful packages to load in \n",
    "\n",
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import matplotlib.pyplot as plt # data visualisation\n",
    "import tensorflow as tf\n",
    "\n",
    "\n",
    "# Input data files are available in the \"../input/\" directory.\n",
    "# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n",
    "\n",
    "import os\n",
    "print(os.listdir(\"../input\"))\n",
    "\n",
    "# Any results you write to the current directory are saved as output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
    "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a"
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv(\"../input/year_prediction.csv\")\n",
    "data = data.rename(index=str, columns={\"label\":\"year\"})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We plot a histogram to understand how evenly spread the data is by viewing number of songs we have for a given year."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Number of songs')"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZUAAAEKCAYAAADaa8itAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAHopJREFUeJzt3X2UHVWZ7/Hvj/AiKpgALSvmxY4a566Iw4s9kBnn+gLXpAkzBBUY8IUszDXOMtyFV2Yk8aqoiAYdweEKzGSGXMOMGhh8oYVoJmLU0SUhDcSQBJg0IUoygUQSSJBFMOG5f9TupNI5p7uSVJ2T0/37rFWrq57aVbWrOPBQtXftUkRgZmZWhsOaXQEzMxs8nFTMzKw0TipmZlYaJxUzMyuNk4qZmZXGScXMzErjpGJmZqVxUjEzs9I4qZiZWWkOb3YFGu2EE06I9vb2ZlfDzKxlnHDCCSxatGhRRHQOVHbIJZX29na6u7ubXQ0zs5Yi6YQi5fz4y8zMSuOkYmZmpak8qUgaJulBSXel5XGSlkrqkXSbpCNT/Ki03JPWt+f2MTvFH5U0ORfvTLEeSbOqPhczM+tfI+5ULgcezi1fC1wfEW8AtgLTU3w6sDXFr0/lkDQBuAh4E9AJ3JQS1TDgRuBsYAJwcSprZmZNUmlSkTQaOAf457Qs4EzgjlRkPnBemp+alknrz0rlpwILImJHRDwO9ACnp6knItZGxIvAglTWzMyapOo7la8BnwBeSsvHA89ExM60vB4YleZHAU8ApPXPpvK74322qRc3M7MmqSypSPoLYFNE3F/VMfajLjMkdUvq3rx5c7OrY2Y2aFV5p/JW4FxJ68geTZ0J/D0wXFLv+zGjgQ1pfgMwBiCtfxXwdD7eZ5t68X1ExNyI6IiIjra2toM/MzMzq6mypBIRsyNidES0kzW0/yQi3g8sAc5PxaYBd6b5rrRMWv+TiIgUvyj1DhsHjAfuA5YB41NvsiPTMbqqOh8zMxtYM96ovxJYIOkLwIPALSl+C/AvknqALWRJgohYJel2YDWwE5gZEbsAJF0GLAKGAfMiYlVDz8TMrAnaZ929e37dnHOaWJN9NSSpRMRPgZ+m+bVkPbf6lnkBuKDO9tcA19SILwQWllhVMzM7CH6j3szMSuOkYmZmpXFSMTOz0jipmJlZaZxUzMysNE4qZmZWGicVMzMrjZOKmZmVZsh9o97MrBXl36I/lPlOxczMSuOkYmZmpXFSMTOz0jipmJlZaZxUzMysNE4qZmZWGicVMzMrjZOKmZmVprKkIullku6T9GtJqyR9LsW/IelxScvTdEqKS9INknokrZB0Wm5f0yStSdO0XPwtkh5K29wgSVWdj5mZDazKN+p3AGdGxHOSjgB+IemHad3fRsQdfcqfDYxP0xnAzcAZko4DrgI6gADul9QVEVtTmQ8DS8k+K9wJ/BAzM2uKyu5UIvNcWjwiTdHPJlOBW9N29wLDJY0EJgOLI2JLSiSLgc607tiIuDciArgVOK+q8zEzs4FV2qYiaZik5cAmssSwNK26Jj3iul7SUSk2Cngit/n6FOsvvr5G3MzMmqTSASUjYhdwiqThwPcknQTMBp4EjgTmAlcCn6+yHpJmADMAxo4dW+WhzMxKkR9Act2cc5pYk/3TkN5fEfEMsATojIiN6RHXDuD/AaenYhuAMbnNRqdYf/HRNeK1jj83IjoioqOtra2MUzIzO2S0z7r7kBnFuMreX23pDgVJRwPvAh5JbSGknlrnASvTJl3AJakX2ETg2YjYCCwCJkkaIWkEMAlYlNZtkzQx7esS4M6qzsfMzAZW5eOvkcB8ScPIktftEXGXpJ9IagMELAf+OpVfCEwBeoDngUsBImKLpKuBZanc5yNiS5r/KPAN4GiyXl/u+WVm1kSVJZWIWAGcWiN+Zp3yAcyss24eMK9GvBs46eBqamZmZfEb9WZmVhonFTMzK42TipmZlcZJxczMSuOkYmZmpXFSMTOz0jipmJlZaZxUzMysNE4qZmZWGicVMzMrjZOKmZmVxknFzMxK46RiZmalcVIxM7PSOKmYmVlpnFTMzKw0VX750czMGiz/rfp1c85p+PGr/Eb9yyTdJ+nXklZJ+lyKj5O0VFKPpNskHZniR6XlnrS+Pbev2Sn+qKTJuXhnivVImlXVuZiZWTFVPv7aAZwZEScDpwCdkiYC1wLXR8QbgK3A9FR+OrA1xa9P5ZA0AbgIeBPQCdwkaZikYcCNwNnABODiVNbMzJqksqQSmefS4hFpCuBM4I4Unw+cl+anpmXS+rMkKcUXRMSOiHgc6AFOT1NPRKyNiBeBBamsmZk1SaUN9emOYjmwCVgMPAY8ExE7U5H1wKg0Pwp4AiCtfxY4Ph/vs029uJmZNUmlSSUidkXEKcBosjuL/1bl8eqRNENSt6TuzZs3N6MKZmZDQkO6FEfEM8AS4E+B4ZJ6e52NBjak+Q3AGIC0/lXA0/l4n23qxWsdf25EdERER1tbWynnZGZm+6qsS7GkNuAPEfGMpKOBd5E1vi8BzidrA5kG3Jk26UrLv0rrfxIRIakL+Jak64DXAOOB+wAB4yWNI0smFwHvq+p8zMyqlu8O3KqqfE9lJDA/9dI6DLg9Iu6StBpYIOkLwIPALan8LcC/SOoBtpAlCSJilaTbgdXATmBmROwCkHQZsAgYBsyLiFUVno+ZmQ2gsqQSESuAU2vE15K1r/SNvwBcUGdf1wDX1IgvBBYedGXNzKwUHqbFzMxK46RiZmalcVIxM7PSDJhUJL1V0ivS/AckXSfptdVXzczMWk2RO5WbgeclnQxcQfZW/K2V1srMzFpSkaSyMyKCbFytr0fEjcAx1VbLzMxaUZEuxdslzQY+ALxN0mFkg0OamZntpcidyl+RDWM/PSKeJBsO5SuV1srMzFrSgHcqKZFcl1v+LW5TMTOzGgZMKpK2k30HJe9ZoBu4Ir0hb2ZmB6B3vK9mfPq3CkXaVL5G9q2Sb5EN4ngR8HrgAWAe8I6qKmdmZq2lSJvKuRHxjxGxPSK2RcRcYHJE3AaMqLh+ZmbWQookleclXSjpsDRdCLyQ1vV9LGZmZkNYkaTyfuCDZJ8EfirNfyB9I+WyCutmZmYtpkjvr7XAX9ZZ/Ytyq2NmZq2sSO+vNuDDQHu+fER8qLpqmZlZKyrS++tO4D+AHwO7qq2OmZm1siJJ5eURcWXlNTEzs5ZXpKH+LklT9nfHksZIWiJptaRVki5P8c9K2iBpeZqm5LaZLalH0qOSJufinSnWI2lWLj5O0tIUv03SkftbTzMzK0+RpHI5WWJ5QdL2NG0rsN1OsjfuJwATgZmSJqR110fEKWlaCJDWXQS8CegEbpI0TNIw4EbgbGACcHFuP9emfb0B2ApML3TWZmZWiQGTSkQcExGHRcTL0vwxEXFsge02RsQDaX478DAwqp9NpgILImJHRDwO9ACnp6knItZGxIvAAmCqJAFnAnek7ecD5w1ULzMzq06hzwlLOlfS36XpL/b3IJLagVOBpSl0maQVkuZJ6n0rfxTwRG6z9SlWL3488ExE7OwTr3X8GZK6JXVv3rx5f6tvZmYFFfmc8ByyR2Cr03S5pC8VPYCkVwLfAT4WEdvIviT5euAUYCPw1QOo936JiLkR0RERHW1tbVUfzsxsyCrS+2sKcEpEvAQgaT7wIDB7oA0lHUGWUL4ZEd8FiIincuv/CbgrLW4AxuQ2H51i1Ik/DQyXdHi6W8mXNzOzJij0+AsYnpt/VZENUpvHLcDDEXFdLj4yV+zdwMo03wVcJOkoSeOA8cB9wDJgfOrpdSRZY35X+sTxEuD8tP00sndqzMysSYrcqXwJeFDSErKh798GzOp/EwDeSjZO2EOSlqfYJ8l6b51CNhjlOuAjABGxStLtZI/YdgIzI2IXgKTLgEXAMGBeRKxK+7sSWCDpC2R3T7cUqJeZmVWkyNhf35b0U+BPUujK9DXIgbb7BVkS6mthP9tcA1xTI76w1nZpXLLTB6qLmZk1RpGG+rcC2yKiCzgW+ISk11ZeMzMzazlF2lRuJvumysnAx4HH8DfqzcyshiJJZWdqFJ8K3BgRNwLHVFstMzNrRUUa6rdLmg18AHibpMOAI6qtlpnZ4NQ+6+7d8+vmnNPEmlSjyJ3KXwE7gOmpgX408JVKa2VmZi2pSO+vJ4Hrcsu/xW0qZmZWQ9GXH83MzAbkpGJmZqWpm1Qk3ZP+Xtu46piZWSvrr01lpKQ/A86VtIA+b8f3fivFzMysV39J5TPAp8l6e13XZ12QfSDLzMxst7pJJSLuAO6Q9OmIuLqBdTIzsxZVpEvx1ZLOJRudGOCnEXFXf9uYmdkevS88DsaXHfsqMqDkl9j3y49frLpiZmbWeooM03IOtb/8+MkqK2ZmZq2nsi8/mpnZ0FPllx/NzGyIGfBOJSK+DUwEvgt8B/jTiLhtoO0kjZG0RNJqSaskXZ7ix0laLGlN+jsixSXpBkk9klZIOi23r2mp/BpJ03Lxt0h6KG1zg6RaX5o0M7MGKfT4KyI2RkRXmgb8lHCyE7giIiaQJaWZkiaQ3eXcExHjgXvYc9dzNjA+TTPIPg6GpOOAq4AzyD4dfFVvIkplPpzbrrNg3czMrAKVjf2VEtEDaX478DAwiuxjX/NTsfnAeWl+KnBrZO4FhksaCUwGFkfElojYCiwGOtO6YyPi3vQRsVtz+zIzsyZoyICSktqBU4GlwIkRsTGtehI4Mc2PAp7IbbY+xfqLr68RNzOzJuk3qUgaJumRgzmApFeStcV8LCK25delO4w4mP0XrMMMSd2Sujdv3lz14czMhqx+k0pE7AIelTT2QHYu6QiyhPLNiPhuCj+VHl2R/m5K8Q3AmNzmo1Osv/joGvFa5zE3IjoioqOtre1ATsXMzAoo8vhrBLBK0j2SunqngTZKPbFuAR6OiPyAlF1Abw+uacCdufglqRfYRODZ9JhsETBJ0ojUQD8JWJTWbZM0MR3rkty+zMysCYq8p/LpA9z3W4EPAg9JWp5inwTmALdLmg78BrgwrVsITAF6gOeBSwEiYoukq4FlqdznI2JLmv8o8A3gaOCHaTIza6resb5gaIz3lVdkQMmfSXotMD4ifizp5cCwAtv9gj7fYMk5q0b5AGbW2dc8YF6NeDdw0kB1MTOzxigyoOSHgTuAf0yhUcD3q6yUmZm1piJtKjPJHmVtA4iINcCrq6yUmZm1piJJZUdEvNi7IOlwGtAN2MzMWk+RpPIzSZ8Ejpb0LuDfgB9UWy0zM2tFRXp/zQKmAw8BHyHrpfXPVVbKzKzV5Ht8DWVFen+9lD7MtZTssdejqaeWmZnZXgZMKpLOAf4BeIysi/A4SR+JCL8TYmZmeyny+OurwDsjogdA0uuBu/GLhmZm1keRhvrtvQklWQtsr6g+ZmbWwureqUh6T5rtlrQQuJ2sTeUC9gyZYmZmtlt/j7/+Mjf/FPD2NL+ZbKwtMzOzvdRNKhFxaSMrYmZmra9I769xwP8C2vPlI+Lc6qplZmatqEjvr++TfRflB8BL1VbHzMxaWZGk8kJE3FB5TczMrOUVSSp/L+kq4N+BHb3BiHigslqZmVlLKpJU3kz2Bccz2fP4K9KymZnZbkVefrwAeF1EvD0i3pmmAROKpHmSNklamYt9VtIGScvTNCW3brakHkmPSpqci3emWI+kWbn4OElLU/w2SUcWP20zs4PTPuvu3ZPtUSSprASGH8C+vwF01ohfHxGnpGkhgKQJwEXAm9I2N0kaJmkYcCNwNjABuDiVBbg27esNwFaykZTNzKyJijz+Gg48ImkZe7ep9NulOCJ+Lqm9YD2mAgsiYgfwuKQe4PS0rici1gJIWgBMlfQw2eO396Uy84HPAjcXPJ6ZmVWgSFK5quRjXibpEqAbuCIitpJ99/7eXJn1KQbwRJ/4GcDxwDMRsbNGeTOzSvhR18CKfE/lZyUe72bgarKG/qvJRkD+UIn7r0nSDGAGwNixY6s+nJnZkDVgm4qk7ZK2pekFSbskbTuQg0XEUxGxKyJeAv6JPY+4NgBjckVHp1i9+NPAcEmH94nXO+7ciOiIiI62trYDqbqZmRUwYFKJiGMi4tiIOJZsIMn3AjcdyMEkjcwtvpusEwBAF3CRpKPSsDDjgfvIRkMen3p6HUnWmN+Vvjy5BDg/bT8NuPNA6mRmZuUp0vtrt8h8H5g8UFlJ3wZ+BfyRpPWSpgNflvSQpBXAO4H/nfa7imxo/dXAj4CZ6Y5mJ3AZsAh4GLg9lQW4Evh4atQ/nmwoGTMza6IiA0q+J7d4GNABvDDQdhFxcY1w3f/wR8Q1wDU14guBhTXia9nz+MzMzA4BRXp/5b+rshNYR9YF2Mxs0Ovt8bVuzjlNrklrKNL7y99VMTOzQvr7nPBn+tkuIuLqCupjZmYtrL87ld/XiL2CbDiU48neMzEzM9utv88Jf7V3XtIxwOXApcACspcWzcwGnfxb825H2X/9tqlIOg74OPB+svG1TkvDqpiZme2jvzaVrwDvAeYCb46I5xpWKzMza0n9vfx4BfAa4FPAf+WGatl+oMO0mJnZ4NZfm8p+vW1vZmbmxGFmZqVxUjGzIc+fBS6Pk4qZmZXGScXMzErjpGJmZqVxUjEzs9I4qZiZWWmcVMzMrDROKmZmVprKkoqkeZI2SVqZix0nabGkNenviBSXpBsk9UhaIem03DbTUvk1kqbl4m9J37vvSduqqnMxs8Gl970Uv5tSvirvVL4BdPaJzQLuiYjxwD1pGeBsYHyaZgA3w+5Rkq8CziD7Hv1VvYkolflwbru+xzIzswarLKlExM+BLX3CU8mG0Cf9PS8XvzUy9wLDJY0EJgOLI2JLGnJ/MdCZ1h0bEfdGRAC35vZlZmZN0ug2lRMjYmOafxI4Mc2PAp7IlVufYv3F19eI1yRphqRuSd2bN28+uDMwM7O6mtZQn+4wokHHmhsRHRHR0dbW1ohDmpkNSY1OKk+lR1ekv5tSfAMwJldudIr1Fx9dI25mZk3U7+eEK9AFTAPmpL935uKXSVpA1ij/bERslLQI+GKucX4SMDsitqQPhk0ElgKXAP+3kSdiZq1lqPb06j3vdXPOacjxKksqkr4NvAM4QdJ6sl5cc4DbJU0HfgNcmIovBKYAPcDzwKUAKXlcDSxL5T4fEb2N/x8l62F2NPDDNJmZWRNVllQi4uI6q86qUTaAmXX2Mw+YVyPeDZx0MHU0M7Ny+Y16MzMrTaPbVMzMGiLfhtKo9gTznYqZmZXIScXMzErjpGJmZqVxUjEzs9I4qZiZWWmcVMzMrDROKmZmVhq/p2Jmg8ZQHd/rUOI7FTMzK42TipmZlcZJxcxaWvusu/3Y6xDipGJmZqVxUjEzs9I4qZiZWWmcVMzMrDRNSSqS1kl6SNJySd0pdpykxZLWpL8jUlySbpDUI2mFpNNy+5mWyq+RNK0Z52JmjdXbMO/G+UNTM+9U3hkRp0RER1qeBdwTEeOBe9IywNnA+DTNAG6GLAmRfff+DOB04KreRGRmZs1xKD3+mgrMT/PzgfNy8Vsjcy8wXNJIYDKwOCK2RMRWYDHQ2ehKm5nZHs1KKgH8u6T7Jc1IsRMjYmOafxI4Mc2PAp7Ibbs+xerFzWyQ8eOu1tGssb/+PCI2SHo1sFjSI/mVERGSoqyDpcQ1A2Ds2LFl7dbMzPpoyp1KRGxIfzcB3yNrE3kqPdYi/d2Uim8AxuQ2H51i9eK1jjc3IjoioqOtra3MUzEzs5yGJxVJr5B0TO88MAlYCXQBvT24pgF3pvku4JLUC2wi8Gx6TLYImCRpRGqgn5RiZmbWJM14/HUi8D1Jvcf/VkT8SNIy4HZJ04HfABem8guBKUAP8DxwKUBEbJF0NbAslft8RGxp3GmYmVlfDU8qEbEWOLlG/GngrBrxAGbW2dc8YF7ZdTSz5so3yq+bc04Ta2L761DqUmxmZi3OScXMzErjzwmbWaV6H2UN9BjL76EMDk4qZtYw9RKH200GDycVM9tv/TWku5F9aHNSMbNC+ns8VfQRlw1+bqg3M7PSOKmYDQEekNEaxY+/zFpUWW0X9R5duW3EDoSTilkLKXq3caAN6b6bsYPlpGI2SDgh2KHAbSpmhzi3h1gr8Z2K2SHGbRnWynynYmZmpXFSMTsE+BGXDRZ+/GXWBH7EZYOVk4pZRfomDt+J2FDQ8o+/JHVKelRSj6RZza6PmdlQ1tJ3KpKGATcC7wLWA8skdUXE6ubWzFpB0TuJ/Lpaj6ry63w3YkNdSycV4HSgJ333HkkLgKmAk4rVVMZ/9N0eYlZfqyeVUcATueX1wBlNqstB83+s9nagdxL1tvM1NaueIqLZdThgks4HOiPif6blDwJnRMRlfcrNAGakxT8CHh1g1ycAvyu5uq3O12RfviZ78/XY12C5Jr8DiIjOgQq2+p3KBmBMbnl0iu0lIuYCc4vuVFJ3RHQcfPUGD1+Tffma7M3XY19D8Zq0eu+vZcB4SeMkHQlcBHQ1uU5mZkNWS9+pRMROSZcBi4BhwLyIWNXkapmZDVktnVQAImIhsLDk3RZ+VDaE+Jrsy9dkb74e+xpy16SlG+rNzOzQ0uptKmZmdggZMklF0jxJmyStzMVOlvQrSQ9J+oGkY1P8XZLuT/H7JZ2Z2+YtKd4j6QZJasb5HKz9uR659WMlPSfpb3KxQTNMzv5eE0l/nNatSutfluKD4jcC+/3vzRGS5qf4w5Jm57YZFL8TSWMkLZG0Ov1zvzzFj5O0WNKa9HdEiiv9BnokrZB0Wm5f01L5NZKmNeucShcRQ2IC3gacBqzMxZYBb0/zHwKuTvOnAq9J8ycBG3Lb3AdMBAT8EDi72edW9fXIrb8D+Dfgb9LyMOAx4HXAkcCvgQnNPrcG/UYOB1YAJ6fl44Fhg+k3cgDX5H3AgjT/cmAd0D6YfifASOC0NH8M8J/ABODLwKwUnwVcm+anpN+A0m9iaYofB6xNf0ek+RHNPr8ypiFzpxIRPwe29Am/Efh5ml8MvDeVfTAi/ivFVwFHSzpK0kjg2Ii4N7Jfxq3AedXXvnz7cz0AJJ0HPE52PXrtHiYnIl4EeofJaUn7eU0mASsi4tdp26cjYtdg+o3Afl+TAF4h6XDgaOBFYBuD6HcSERsj4oE0vx14mGxkj6nA/FRsPnv+mU8Fbo3MvcDw9BuZDCyOiC0RsZXsOg74YmErGDJJpY5V7PlxX8DeL1L2ei/wQETsIPvxrM+tW59ig0XN6yHplcCVwOf6lK81TM5guh5Q/zfyRiAkLZL0gKRPpPhg/41A/WtyB/B7YCPwW+DvImILg/R3Iqmd7KnGUuDEiNiYVj0JnJjm6537oLwm4KTyIeCjku4nu5V9Mb9S0puAa4GPNKFuzVDvenwWuD4inmtWxZqo3jU5HPhz4P3p77slndWcKjZcvWtyOrALeA0wDrhC0uuaU8Vqpf/R+g7wsYjYll+X7lCHbLfaln9P5WBExCNkjzGQ9EZg94iDkkYD3wMuiYjHUngD2VAwvWoOC9Oq+rkeZwDnS/oyMBx4SdILwP0UGCanlfVzTdYDP4+I36V1C8naHv6VQfwbgX6vyfuAH0XEH4BNkn4JdJD9H/mg+Z1IOoIsoXwzIr6bwk9JGhkRG9PjrU0pXm8oqQ3AO/rEf1plvRtlSN+pSHp1+nsY8CngH9LycOBusoa3X/aWT7e32yRNTD16LgHubHjFK1LvekTEf4+I9ohoB74GfDEivs4QGCan3jUhG8XhzZJentoQ3g6sHuy/Eej3mvwWODOtewVZw/QjDKLfSfpnegvwcERcl1vVBfT24JrGnn/mXcAlqRfYRODZ9BtZBEySNCL1FJuUYq2v2T0FGjUB3yZ71vsHsv/LnA5cTtZ74z+BOex5GfRTZM+Gl+emV6d1HcBKst4sX+/dptWm/bkefbb7LKn3V1qekso/BvyfZp9XI68J8AGy9oWVwJdz8UHxG9nfawK8kqx34Cqybxr97WD7nZA96gyynn+9/22YQtb77x5gDfBj4LhUXmQfEnwMeAjoyO3rQ0BPmi5t9rmVNfmNejMzK82QfvxlZmblclIxM7PSOKmYmVlpnFTMzKw0TipmZlYaJxWzCqX3E34h6exc7AJJP2pmvcyq4i7FZhWTdBLZ+xunko1i8SDQGXtGajiQfR4eETtLqqJZaXynYlaxiFgJ/IBsUM7PkI1a+1j6nsZ9kpZLuim9oY6kuZK60/c6PtO7H0nrJc2R9CDw7qacjNkAhvTYX2YN9DngAbLBFzvS3cu7gT+LiJ2S5pINX/ItsuGBtqThX5ZIuiMiVqf9bIqIU5txAmZFOKmYNUBE/F7SbcBzEbFD0v8A/gTozoaT4mj2DIV+saTpZP9+vobsI1C9SeW2xtbcbP84qZg1zktpgmxMqHkR8el8AUnjycbWOj0inpH0r8DLckV+35Camh0gt6mYNcePgQslnQAg6XhJY4Fjge1kIx33fiHQrGX4TsWsCSLiIUmfA36cGuj/APw10E32qOsR4DfAL+vvxezQ4y7FZmZWGj/+MjOz0jipmJlZaZxUzMysNE4qZmZWGicVMzMrjZOKmZmVxknFzMxK46RiZmal+f91KJAY/Cg96wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nsongs = {}\n",
    "for y in range(1922,2012):\n",
    "    nsongs[y] = len(data[data.year==y])\n",
    "yrs = range(1922,2011)\n",
    "values = [nsongs[y] for y in yrs]\n",
    "plt.bar(yrs, values, align='center')\n",
    "plt.xlabel(\"Year\")\n",
    "plt.ylabel(\"Number of songs\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# separate input attributes and output into different dataframes\n",
    "X = data.iloc[:,1:]\n",
    "Y = data.iloc[:,0]\n",
    "\n",
    "# Train set\n",
    "X_train = X.iloc[0:463715,:]\n",
    "y_train = Y.iloc[0:463715]\n",
    "\n",
    "# Validation set\n",
    "X_test = X.iloc[463715:,:]\n",
    "y_test = Y.iloc[463715:]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each of the features takes a wide range of different values and distributions.\n",
    "\n",
    "We apply MinMax scaling to our data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scaler = MinMaxScaler()\n",
    "\n",
    "# Fit on training set only.\n",
    "scaler.fit(X_train)\n",
    "# Apply transform to both the train set and the test set.\n",
    "X_train_scaled = scaler.transform(X_train)\n",
    "X_test_scaled = scaler.transform(X_test)\n",
    "\n",
    "X_train = pd.DataFrame(X_train_scaled,columns=X_train.columns)\n",
    "X_test = pd.DataFrame(X_test_scaled,columns=X_train.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>TimbreAvg1</th>\n",
       "      <th>TimbreAvg2</th>\n",
       "      <th>TimbreAvg3</th>\n",
       "      <th>TimbreAvg4</th>\n",
       "      <th>TimbreAvg5</th>\n",
       "      <th>TimbreAvg6</th>\n",
       "      <th>TimbreAvg7</th>\n",
       "      <th>TimbreAvg8</th>\n",
       "      <th>TimbreAvg9</th>\n",
       "      <th>TimbreAvg10</th>\n",
       "      <th>TimbreAvg11</th>\n",
       "      <th>TimbreAvg12</th>\n",
       "      <th>TimbreCovariance1</th>\n",
       "      <th>TimbreCovariance2</th>\n",
       "      <th>TimbreCovariance3</th>\n",
       "      <th>TimbreCovariance4</th>\n",
       "      <th>TimbreCovariance5</th>\n",
       "      <th>TimbreCovariance6</th>\n",
       "      <th>TimbreCovariance7</th>\n",
       "      <th>TimbreCovariance8</th>\n",
       "      <th>TimbreCovariance9</th>\n",
       "      <th>TimbreCovariance10</th>\n",
       "      <th>TimbreCovariance11</th>\n",
       "      <th>TimbreCovariance12</th>\n",
       "      <th>TimbreCovariance13</th>\n",
       "      <th>TimbreCovariance14</th>\n",
       "      <th>TimbreCovariance15</th>\n",
       "      <th>TimbreCovariance16</th>\n",
       "      <th>TimbreCovariance17</th>\n",
       "      <th>TimbreCovariance18</th>\n",
       "      <th>TimbreCovariance19</th>\n",
       "      <th>TimbreCovariance20</th>\n",
       "      <th>TimbreCovariance21</th>\n",
       "      <th>TimbreCovariance22</th>\n",
       "      <th>TimbreCovariance23</th>\n",
       "      <th>TimbreCovariance24</th>\n",
       "      <th>TimbreCovariance25</th>\n",
       "      <th>TimbreCovariance26</th>\n",
       "      <th>TimbreCovariance27</th>\n",
       "      <th>TimbreCovariance28</th>\n",
       "      <th>...</th>\n",
       "      <th>TimbreCovariance39</th>\n",
       "      <th>TimbreCovariance40</th>\n",
       "      <th>TimbreCovariance41</th>\n",
       "      <th>TimbreCovariance42</th>\n",
       "      <th>TimbreCovariance43</th>\n",
       "      <th>TimbreCovariance44</th>\n",
       "      <th>TimbreCovariance45</th>\n",
       "      <th>TimbreCovariance46</th>\n",
       "      <th>TimbreCovariance47</th>\n",
       "      <th>TimbreCovariance48</th>\n",
       "      <th>TimbreCovariance49</th>\n",
       "      <th>TimbreCovariance50</th>\n",
       "      <th>TimbreCovariance51</th>\n",
       "      <th>TimbreCovariance52</th>\n",
       "      <th>TimbreCovariance53</th>\n",
       "      <th>TimbreCovariance54</th>\n",
       "      <th>TimbreCovariance55</th>\n",
       "      <th>TimbreCovariance56</th>\n",
       "      <th>TimbreCovariance57</th>\n",
       "      <th>TimbreCovariance58</th>\n",
       "      <th>TimbreCovariance59</th>\n",
       "      <th>TimbreCovariance60</th>\n",
       "      <th>TimbreCovariance61</th>\n",
       "      <th>TimbreCovariance62</th>\n",
       "      <th>TimbreCovariance63</th>\n",
       "      <th>TimbreCovariance64</th>\n",
       "      <th>TimbreCovariance65</th>\n",
       "      <th>TimbreCovariance66</th>\n",
       "      <th>TimbreCovariance67</th>\n",
       "      <th>TimbreCovariance68</th>\n",
       "      <th>TimbreCovariance69</th>\n",
       "      <th>TimbreCovariance70</th>\n",
       "      <th>TimbreCovariance71</th>\n",
       "      <th>TimbreCovariance72</th>\n",
       "      <th>TimbreCovariance73</th>\n",
       "      <th>TimbreCovariance74</th>\n",
       "      <th>TimbreCovariance75</th>\n",
       "      <th>TimbreCovariance76</th>\n",
       "      <th>TimbreCovariance77</th>\n",
       "      <th>TimbreCovariance78</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "      <td>463715.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.691393</td>\n",
       "      <td>0.469181</td>\n",
       "      <td>0.496357</td>\n",
       "      <td>0.350035</td>\n",
       "      <td>0.395117</td>\n",
       "      <td>0.358260</td>\n",
       "      <td>0.515310</td>\n",
       "      <td>0.397887</td>\n",
       "      <td>0.477290</td>\n",
       "      <td>0.426607</td>\n",
       "      <td>0.691819</td>\n",
       "      <td>0.530809</td>\n",
       "      <td>0.061015</td>\n",
       "      <td>0.036932</td>\n",
       "      <td>0.052829</td>\n",
       "      <td>0.046941</td>\n",
       "      <td>0.045245</td>\n",
       "      <td>0.051828</td>\n",
       "      <td>0.054789</td>\n",
       "      <td>0.053319</td>\n",
       "      <td>0.040334</td>\n",
       "      <td>0.083562</td>\n",
       "      <td>0.041914</td>\n",
       "      <td>0.029258</td>\n",
       "      <td>0.588071</td>\n",
       "      <td>0.354463</td>\n",
       "      <td>0.451344</td>\n",
       "      <td>0.555480</td>\n",
       "      <td>0.508205</td>\n",
       "      <td>0.473871</td>\n",
       "      <td>0.676466</td>\n",
       "      <td>0.443812</td>\n",
       "      <td>0.369822</td>\n",
       "      <td>0.404615</td>\n",
       "      <td>0.353309</td>\n",
       "      <td>0.424043</td>\n",
       "      <td>0.371572</td>\n",
       "      <td>0.562552</td>\n",
       "      <td>0.605844</td>\n",
       "      <td>0.486739</td>\n",
       "      <td>...</td>\n",
       "      <td>0.615812</td>\n",
       "      <td>0.421879</td>\n",
       "      <td>0.573337</td>\n",
       "      <td>0.431647</td>\n",
       "      <td>0.744069</td>\n",
       "      <td>0.400336</td>\n",
       "      <td>0.372550</td>\n",
       "      <td>0.266772</td>\n",
       "      <td>0.362311</td>\n",
       "      <td>0.301800</td>\n",
       "      <td>0.512831</td>\n",
       "      <td>0.537379</td>\n",
       "      <td>0.417522</td>\n",
       "      <td>0.477130</td>\n",
       "      <td>0.719306</td>\n",
       "      <td>0.530612</td>\n",
       "      <td>0.369416</td>\n",
       "      <td>0.353011</td>\n",
       "      <td>0.269319</td>\n",
       "      <td>0.554657</td>\n",
       "      <td>0.639893</td>\n",
       "      <td>0.595036</td>\n",
       "      <td>0.795590</td>\n",
       "      <td>0.521606</td>\n",
       "      <td>0.512374</td>\n",
       "      <td>0.488463</td>\n",
       "      <td>0.494068</td>\n",
       "      <td>0.376795</td>\n",
       "      <td>0.631798</td>\n",
       "      <td>0.563245</td>\n",
       "      <td>0.354631</td>\n",
       "      <td>0.487984</td>\n",
       "      <td>0.368845</td>\n",
       "      <td>0.657332</td>\n",
       "      <td>0.568014</td>\n",
       "      <td>0.466481</td>\n",
       "      <td>0.600205</td>\n",
       "      <td>0.343815</td>\n",
       "      <td>0.503517</td>\n",
       "      <td>0.347715</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.100947</td>\n",
       "      <td>0.071570</td>\n",
       "      <td>0.056527</td>\n",
       "      <td>0.036814</td>\n",
       "      <td>0.051474</td>\n",
       "      <td>0.063671</td>\n",
       "      <td>0.040431</td>\n",
       "      <td>0.044802</td>\n",
       "      <td>0.038783</td>\n",
       "      <td>0.063964</td>\n",
       "      <td>0.043463</td>\n",
       "      <td>0.045744</td>\n",
       "      <td>0.040440</td>\n",
       "      <td>0.026595</td>\n",
       "      <td>0.034290</td>\n",
       "      <td>0.034337</td>\n",
       "      <td>0.023954</td>\n",
       "      <td>0.034274</td>\n",
       "      <td>0.029750</td>\n",
       "      <td>0.032271</td>\n",
       "      <td>0.022275</td>\n",
       "      <td>0.044558</td>\n",
       "      <td>0.027650</td>\n",
       "      <td>0.015673</td>\n",
       "      <td>0.024969</td>\n",
       "      <td>0.018895</td>\n",
       "      <td>0.020548</td>\n",
       "      <td>0.028204</td>\n",
       "      <td>0.024609</td>\n",
       "      <td>0.020223</td>\n",
       "      <td>0.017601</td>\n",
       "      <td>0.020666</td>\n",
       "      <td>0.016301</td>\n",
       "      <td>0.013634</td>\n",
       "      <td>0.015131</td>\n",
       "      <td>0.025654</td>\n",
       "      <td>0.018390</td>\n",
       "      <td>0.024946</td>\n",
       "      <td>0.016436</td>\n",
       "      <td>0.012650</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025309</td>\n",
       "      <td>0.017857</td>\n",
       "      <td>0.012536</td>\n",
       "      <td>0.016366</td>\n",
       "      <td>0.025507</td>\n",
       "      <td>0.023793</td>\n",
       "      <td>0.022147</td>\n",
       "      <td>0.015102</td>\n",
       "      <td>0.020012</td>\n",
       "      <td>0.017587</td>\n",
       "      <td>0.015973</td>\n",
       "      <td>0.023356</td>\n",
       "      <td>0.026971</td>\n",
       "      <td>0.014372</td>\n",
       "      <td>0.021669</td>\n",
       "      <td>0.017463</td>\n",
       "      <td>0.030666</td>\n",
       "      <td>0.015855</td>\n",
       "      <td>0.014724</td>\n",
       "      <td>0.021925</td>\n",
       "      <td>0.022821</td>\n",
       "      <td>0.019606</td>\n",
       "      <td>0.015579</td>\n",
       "      <td>0.018304</td>\n",
       "      <td>0.024269</td>\n",
       "      <td>0.018045</td>\n",
       "      <td>0.013502</td>\n",
       "      <td>0.026496</td>\n",
       "      <td>0.015304</td>\n",
       "      <td>0.014604</td>\n",
       "      <td>0.025093</td>\n",
       "      <td>0.019768</td>\n",
       "      <td>0.024255</td>\n",
       "      <td>0.019886</td>\n",
       "      <td>0.026819</td>\n",
       "      <td>0.016738</td>\n",
       "      <td>0.024350</td>\n",
       "      <td>0.019045</td>\n",
       "      <td>0.012446</td>\n",
       "      <td>0.023989</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.634471</td>\n",
       "      <td>0.431166</td>\n",
       "      <td>0.464150</td>\n",
       "      <td>0.328296</td>\n",
       "      <td>0.363307</td>\n",
       "      <td>0.314100</td>\n",
       "      <td>0.492040</td>\n",
       "      <td>0.371622</td>\n",
       "      <td>0.455227</td>\n",
       "      <td>0.384236</td>\n",
       "      <td>0.666902</td>\n",
       "      <td>0.502762</td>\n",
       "      <td>0.032903</td>\n",
       "      <td>0.019860</td>\n",
       "      <td>0.029653</td>\n",
       "      <td>0.024999</td>\n",
       "      <td>0.029936</td>\n",
       "      <td>0.028599</td>\n",
       "      <td>0.034962</td>\n",
       "      <td>0.032599</td>\n",
       "      <td>0.026155</td>\n",
       "      <td>0.053510</td>\n",
       "      <td>0.023966</td>\n",
       "      <td>0.019620</td>\n",
       "      <td>0.576851</td>\n",
       "      <td>0.345704</td>\n",
       "      <td>0.443649</td>\n",
       "      <td>0.543756</td>\n",
       "      <td>0.496901</td>\n",
       "      <td>0.463525</td>\n",
       "      <td>0.668716</td>\n",
       "      <td>0.435032</td>\n",
       "      <td>0.362304</td>\n",
       "      <td>0.398372</td>\n",
       "      <td>0.347068</td>\n",
       "      <td>0.410214</td>\n",
       "      <td>0.363964</td>\n",
       "      <td>0.550613</td>\n",
       "      <td>0.599561</td>\n",
       "      <td>0.481041</td>\n",
       "      <td>...</td>\n",
       "      <td>0.603614</td>\n",
       "      <td>0.412993</td>\n",
       "      <td>0.567825</td>\n",
       "      <td>0.424660</td>\n",
       "      <td>0.735812</td>\n",
       "      <td>0.388816</td>\n",
       "      <td>0.362327</td>\n",
       "      <td>0.260170</td>\n",
       "      <td>0.352446</td>\n",
       "      <td>0.293710</td>\n",
       "      <td>0.505172</td>\n",
       "      <td>0.526958</td>\n",
       "      <td>0.404470</td>\n",
       "      <td>0.471317</td>\n",
       "      <td>0.710898</td>\n",
       "      <td>0.523496</td>\n",
       "      <td>0.354213</td>\n",
       "      <td>0.346601</td>\n",
       "      <td>0.262169</td>\n",
       "      <td>0.545554</td>\n",
       "      <td>0.631992</td>\n",
       "      <td>0.586874</td>\n",
       "      <td>0.789989</td>\n",
       "      <td>0.512644</td>\n",
       "      <td>0.501426</td>\n",
       "      <td>0.481987</td>\n",
       "      <td>0.487710</td>\n",
       "      <td>0.365284</td>\n",
       "      <td>0.627335</td>\n",
       "      <td>0.556783</td>\n",
       "      <td>0.340913</td>\n",
       "      <td>0.480545</td>\n",
       "      <td>0.356429</td>\n",
       "      <td>0.648426</td>\n",
       "      <td>0.556221</td>\n",
       "      <td>0.459283</td>\n",
       "      <td>0.589716</td>\n",
       "      <td>0.333782</td>\n",
       "      <td>0.498175</td>\n",
       "      <td>0.336685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.705958</td>\n",
       "      <td>0.479041</td>\n",
       "      <td>0.499274</td>\n",
       "      <td>0.345928</td>\n",
       "      <td>0.396288</td>\n",
       "      <td>0.350109</td>\n",
       "      <td>0.516244</td>\n",
       "      <td>0.398215</td>\n",
       "      <td>0.477664</td>\n",
       "      <td>0.425654</td>\n",
       "      <td>0.692281</td>\n",
       "      <td>0.529513</td>\n",
       "      <td>0.052644</td>\n",
       "      <td>0.030402</td>\n",
       "      <td>0.045217</td>\n",
       "      <td>0.038357</td>\n",
       "      <td>0.040531</td>\n",
       "      <td>0.043547</td>\n",
       "      <td>0.048959</td>\n",
       "      <td>0.045778</td>\n",
       "      <td>0.035742</td>\n",
       "      <td>0.074719</td>\n",
       "      <td>0.034995</td>\n",
       "      <td>0.026409</td>\n",
       "      <td>0.584910</td>\n",
       "      <td>0.352183</td>\n",
       "      <td>0.451626</td>\n",
       "      <td>0.556034</td>\n",
       "      <td>0.507122</td>\n",
       "      <td>0.471686</td>\n",
       "      <td>0.676283</td>\n",
       "      <td>0.443550</td>\n",
       "      <td>0.369430</td>\n",
       "      <td>0.403198</td>\n",
       "      <td>0.354251</td>\n",
       "      <td>0.420130</td>\n",
       "      <td>0.370622</td>\n",
       "      <td>0.560477</td>\n",
       "      <td>0.607698</td>\n",
       "      <td>0.486623</td>\n",
       "      <td>...</td>\n",
       "      <td>0.615471</td>\n",
       "      <td>0.421269</td>\n",
       "      <td>0.572661</td>\n",
       "      <td>0.431431</td>\n",
       "      <td>0.747703</td>\n",
       "      <td>0.397364</td>\n",
       "      <td>0.371202</td>\n",
       "      <td>0.265863</td>\n",
       "      <td>0.361004</td>\n",
       "      <td>0.301399</td>\n",
       "      <td>0.511876</td>\n",
       "      <td>0.538022</td>\n",
       "      <td>0.415502</td>\n",
       "      <td>0.478171</td>\n",
       "      <td>0.720191</td>\n",
       "      <td>0.531227</td>\n",
       "      <td>0.369155</td>\n",
       "      <td>0.351679</td>\n",
       "      <td>0.267990</td>\n",
       "      <td>0.554698</td>\n",
       "      <td>0.642040</td>\n",
       "      <td>0.595653</td>\n",
       "      <td>0.796303</td>\n",
       "      <td>0.521472</td>\n",
       "      <td>0.514188</td>\n",
       "      <td>0.490602</td>\n",
       "      <td>0.492406</td>\n",
       "      <td>0.378191</td>\n",
       "      <td>0.634349</td>\n",
       "      <td>0.563230</td>\n",
       "      <td>0.349484</td>\n",
       "      <td>0.490259</td>\n",
       "      <td>0.366313</td>\n",
       "      <td>0.656434</td>\n",
       "      <td>0.568816</td>\n",
       "      <td>0.466181</td>\n",
       "      <td>0.600938</td>\n",
       "      <td>0.341908</td>\n",
       "      <td>0.502698</td>\n",
       "      <td>0.346333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.765257</td>\n",
       "      <td>0.517551</td>\n",
       "      <td>0.530164</td>\n",
       "      <td>0.367222</td>\n",
       "      <td>0.427238</td>\n",
       "      <td>0.393690</td>\n",
       "      <td>0.539991</td>\n",
       "      <td>0.424327</td>\n",
       "      <td>0.500152</td>\n",
       "      <td>0.468346</td>\n",
       "      <td>0.717497</td>\n",
       "      <td>0.557309</td>\n",
       "      <td>0.078641</td>\n",
       "      <td>0.046080</td>\n",
       "      <td>0.066969</td>\n",
       "      <td>0.058735</td>\n",
       "      <td>0.054802</td>\n",
       "      <td>0.065766</td>\n",
       "      <td>0.067635</td>\n",
       "      <td>0.064731</td>\n",
       "      <td>0.048880</td>\n",
       "      <td>0.102909</td>\n",
       "      <td>0.051911</td>\n",
       "      <td>0.035122</td>\n",
       "      <td>0.597456</td>\n",
       "      <td>0.360806</td>\n",
       "      <td>0.459407</td>\n",
       "      <td>0.567574</td>\n",
       "      <td>0.518682</td>\n",
       "      <td>0.481778</td>\n",
       "      <td>0.684181</td>\n",
       "      <td>0.452904</td>\n",
       "      <td>0.376703</td>\n",
       "      <td>0.408807</td>\n",
       "      <td>0.360523</td>\n",
       "      <td>0.434532</td>\n",
       "      <td>0.377622</td>\n",
       "      <td>0.572979</td>\n",
       "      <td>0.614005</td>\n",
       "      <td>0.492403</td>\n",
       "      <td>...</td>\n",
       "      <td>0.627763</td>\n",
       "      <td>0.429982</td>\n",
       "      <td>0.578105</td>\n",
       "      <td>0.438619</td>\n",
       "      <td>0.756769</td>\n",
       "      <td>0.408971</td>\n",
       "      <td>0.381147</td>\n",
       "      <td>0.271646</td>\n",
       "      <td>0.370888</td>\n",
       "      <td>0.308952</td>\n",
       "      <td>0.519270</td>\n",
       "      <td>0.548485</td>\n",
       "      <td>0.428473</td>\n",
       "      <td>0.483641</td>\n",
       "      <td>0.729664</td>\n",
       "      <td>0.537933</td>\n",
       "      <td>0.384313</td>\n",
       "      <td>0.357536</td>\n",
       "      <td>0.274986</td>\n",
       "      <td>0.563966</td>\n",
       "      <td>0.650647</td>\n",
       "      <td>0.603455</td>\n",
       "      <td>0.802214</td>\n",
       "      <td>0.530068</td>\n",
       "      <td>0.524763</td>\n",
       "      <td>0.497455</td>\n",
       "      <td>0.498366</td>\n",
       "      <td>0.390441</td>\n",
       "      <td>0.639235</td>\n",
       "      <td>0.570163</td>\n",
       "      <td>0.362847</td>\n",
       "      <td>0.497755</td>\n",
       "      <td>0.378403</td>\n",
       "      <td>0.665694</td>\n",
       "      <td>0.581546</td>\n",
       "      <td>0.473821</td>\n",
       "      <td>0.611202</td>\n",
       "      <td>0.351684</td>\n",
       "      <td>0.507992</td>\n",
       "      <td>0.356798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          TimbreAvg1         ...          TimbreCovariance78\n",
       "count  463715.000000         ...               463715.000000\n",
       "mean        0.691393         ...                    0.347715\n",
       "std         0.100947         ...                    0.023989\n",
       "min         0.000000         ...                    0.000000\n",
       "25%         0.634471         ...                    0.336685\n",
       "50%         0.705958         ...                    0.346333\n",
       "75%         0.765257         ...                    0.356798\n",
       "max         1.000000         ...                    1.000000\n",
       "\n",
       "[8 rows x 90 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "90 features is lot of features and so we attempt dimensionality reduction by implementing PCA."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PCA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we normalise our data using scikits StandardScalar.\n",
    "This is a necessary step for pca:\n",
    "> Principle Component Analysis (PCA) as being a prime example of when normalization is important. In PCA we are interested in the components that maximize the variance. If one component (e.g. human height) varies less than another (e.g. weight) because of their respective scales (meters vs. kilos), PCA might determine that the direction of maximal variance more closely corresponds with the ‘weight’ axis, if those features are not scaled. As a change in height of one meter can be considered much more important than the change in weight of one kilogram, this is clearly incorrect.\n",
    "(https://scikit-learn.org/stable/auto_examples/preprocessing/plot_scaling_importance.html#sphx-glr-auto-examples-preprocessing-plot-scaling-importance-py)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "scaler = StandardScaler()\n",
    "# Fit on training set only.\n",
    "scaler.fit(X_train)\n",
    "# Apply transform to both the train set and the test set.\n",
    "X_train_std = scaler.transform(X_train)\n",
    "X_test_std = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>TimbreAvg1</th>\n",
       "      <th>TimbreAvg2</th>\n",
       "      <th>TimbreAvg3</th>\n",
       "      <th>TimbreAvg4</th>\n",
       "      <th>TimbreAvg5</th>\n",
       "      <th>TimbreAvg6</th>\n",
       "      <th>TimbreAvg7</th>\n",
       "      <th>TimbreAvg8</th>\n",
       "      <th>TimbreAvg9</th>\n",
       "      <th>TimbreAvg10</th>\n",
       "      <th>TimbreAvg11</th>\n",
       "      <th>TimbreAvg12</th>\n",
       "      <th>TimbreCovariance1</th>\n",
       "      <th>TimbreCovariance2</th>\n",
       "      <th>TimbreCovariance3</th>\n",
       "      <th>TimbreCovariance4</th>\n",
       "      <th>TimbreCovariance5</th>\n",
       "      <th>TimbreCovariance6</th>\n",
       "      <th>TimbreCovariance7</th>\n",
       "      <th>TimbreCovariance8</th>\n",
       "      <th>TimbreCovariance9</th>\n",
       "      <th>TimbreCovariance10</th>\n",
       "      <th>TimbreCovariance11</th>\n",
       "      <th>TimbreCovariance12</th>\n",
       "      <th>TimbreCovariance13</th>\n",
       "      <th>TimbreCovariance14</th>\n",
       "      <th>TimbreCovariance15</th>\n",
       "      <th>TimbreCovariance16</th>\n",
       "      <th>TimbreCovariance17</th>\n",
       "      <th>TimbreCovariance18</th>\n",
       "      <th>TimbreCovariance19</th>\n",
       "      <th>TimbreCovariance20</th>\n",
       "      <th>TimbreCovariance21</th>\n",
       "      <th>TimbreCovariance22</th>\n",
       "      <th>TimbreCovariance23</th>\n",
       "      <th>TimbreCovariance24</th>\n",
       "      <th>TimbreCovariance25</th>\n",
       "      <th>TimbreCovariance26</th>\n",
       "      <th>TimbreCovariance27</th>\n",
       "      <th>TimbreCovariance28</th>\n",
       "      <th>...</th>\n",
       "      <th>TimbreCovariance39</th>\n",
       "      <th>TimbreCovariance40</th>\n",
       "      <th>TimbreCovariance41</th>\n",
       "      <th>TimbreCovariance42</th>\n",
       "      <th>TimbreCovariance43</th>\n",
       "      <th>TimbreCovariance44</th>\n",
       "      <th>TimbreCovariance45</th>\n",
       "      <th>TimbreCovariance46</th>\n",
       "      <th>TimbreCovariance47</th>\n",
       "      <th>TimbreCovariance48</th>\n",
       "      <th>TimbreCovariance49</th>\n",
       "      <th>TimbreCovariance50</th>\n",
       "      <th>TimbreCovariance51</th>\n",
       "      <th>TimbreCovariance52</th>\n",
       "      <th>TimbreCovariance53</th>\n",
       "      <th>TimbreCovariance54</th>\n",
       "      <th>TimbreCovariance55</th>\n",
       "      <th>TimbreCovariance56</th>\n",
       "      <th>TimbreCovariance57</th>\n",
       "      <th>TimbreCovariance58</th>\n",
       "      <th>TimbreCovariance59</th>\n",
       "      <th>TimbreCovariance60</th>\n",
       "      <th>TimbreCovariance61</th>\n",
       "      <th>TimbreCovariance62</th>\n",
       "      <th>TimbreCovariance63</th>\n",
       "      <th>TimbreCovariance64</th>\n",
       "      <th>TimbreCovariance65</th>\n",
       "      <th>TimbreCovariance66</th>\n",
       "      <th>TimbreCovariance67</th>\n",
       "      <th>TimbreCovariance68</th>\n",
       "      <th>TimbreCovariance69</th>\n",
       "      <th>TimbreCovariance70</th>\n",
       "      <th>TimbreCovariance71</th>\n",
       "      <th>TimbreCovariance72</th>\n",
       "      <th>TimbreCovariance73</th>\n",
       "      <th>TimbreCovariance74</th>\n",
       "      <th>TimbreCovariance75</th>\n",
       "      <th>TimbreCovariance76</th>\n",
       "      <th>TimbreCovariance77</th>\n",
       "      <th>TimbreCovariance78</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>...</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "      <td>4.637150e+05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>7.860182e-16</td>\n",
       "      <td>-2.091763e-15</td>\n",
       "      <td>1.355798e-16</td>\n",
       "      <td>-1.357427e-15</td>\n",
       "      <td>-8.698030e-16</td>\n",
       "      <td>1.346587e-15</td>\n",
       "      <td>-5.625596e-16</td>\n",
       "      <td>4.602650e-16</td>\n",
       "      <td>-3.256487e-16</td>\n",
       "      <td>1.911129e-15</td>\n",
       "      <td>4.018197e-15</td>\n",
       "      <td>-2.502608e-15</td>\n",
       "      <td>-6.190568e-17</td>\n",
       "      <td>-1.069821e-17</td>\n",
       "      <td>-3.529729e-16</td>\n",
       "      <td>-9.763661e-17</td>\n",
       "      <td>-2.724692e-16</td>\n",
       "      <td>2.504086e-16</td>\n",
       "      <td>8.627746e-16</td>\n",
       "      <td>4.121119e-16</td>\n",
       "      <td>-6.631095e-16</td>\n",
       "      <td>3.394529e-16</td>\n",
       "      <td>-7.586569e-16</td>\n",
       "      <td>-8.156104e-17</td>\n",
       "      <td>-8.968334e-15</td>\n",
       "      <td>-2.391046e-15</td>\n",
       "      <td>6.454967e-15</td>\n",
       "      <td>-2.494976e-15</td>\n",
       "      <td>4.325894e-15</td>\n",
       "      <td>2.140151e-17</td>\n",
       "      <td>5.876518e-15</td>\n",
       "      <td>-1.396214e-15</td>\n",
       "      <td>3.546603e-15</td>\n",
       "      <td>1.237439e-14</td>\n",
       "      <td>1.326574e-14</td>\n",
       "      <td>-1.499921e-15</td>\n",
       "      <td>-2.790937e-15</td>\n",
       "      <td>-4.764826e-16</td>\n",
       "      <td>8.122309e-15</td>\n",
       "      <td>5.200963e-15</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.736401e-16</td>\n",
       "      <td>8.136067e-15</td>\n",
       "      <td>2.407077e-14</td>\n",
       "      <td>5.731586e-15</td>\n",
       "      <td>1.305549e-15</td>\n",
       "      <td>5.385272e-16</td>\n",
       "      <td>1.614794e-15</td>\n",
       "      <td>-3.452024e-15</td>\n",
       "      <td>-1.443823e-15</td>\n",
       "      <td>1.322732e-17</td>\n",
       "      <td>2.803837e-15</td>\n",
       "      <td>8.785695e-15</td>\n",
       "      <td>-9.397954e-16</td>\n",
       "      <td>-6.418378e-15</td>\n",
       "      <td>2.609099e-15</td>\n",
       "      <td>-8.714935e-16</td>\n",
       "      <td>2.895461e-15</td>\n",
       "      <td>-1.027189e-14</td>\n",
       "      <td>3.537959e-16</td>\n",
       "      <td>3.607981e-15</td>\n",
       "      <td>-5.811512e-15</td>\n",
       "      <td>3.722570e-15</td>\n",
       "      <td>-4.277726e-15</td>\n",
       "      <td>5.148444e-15</td>\n",
       "      <td>7.730112e-16</td>\n",
       "      <td>-2.349867e-15</td>\n",
       "      <td>-7.733918e-16</td>\n",
       "      <td>2.570563e-15</td>\n",
       "      <td>7.668392e-15</td>\n",
       "      <td>1.063528e-14</td>\n",
       "      <td>1.832125e-16</td>\n",
       "      <td>8.840391e-16</td>\n",
       "      <td>-4.720778e-15</td>\n",
       "      <td>-1.128972e-15</td>\n",
       "      <td>-4.590983e-15</td>\n",
       "      <td>-8.157310e-15</td>\n",
       "      <td>-2.556776e-15</td>\n",
       "      <td>-1.843250e-15</td>\n",
       "      <td>-7.529481e-15</td>\n",
       "      <td>4.832588e-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "      <td>1.000001e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-6.849083e+00</td>\n",
       "      <td>-6.555536e+00</td>\n",
       "      <td>-8.780881e+00</td>\n",
       "      <td>-9.508272e+00</td>\n",
       "      <td>-7.675981e+00</td>\n",
       "      <td>-5.626720e+00</td>\n",
       "      <td>-1.274530e+01</td>\n",
       "      <td>-8.881098e+00</td>\n",
       "      <td>-1.230653e+01</td>\n",
       "      <td>-6.669490e+00</td>\n",
       "      <td>-1.591731e+01</td>\n",
       "      <td>-1.160387e+01</td>\n",
       "      <td>-1.508788e+00</td>\n",
       "      <td>-1.388673e+00</td>\n",
       "      <td>-1.540662e+00</td>\n",
       "      <td>-1.367069e+00</td>\n",
       "      <td>-1.888816e+00</td>\n",
       "      <td>-1.512189e+00</td>\n",
       "      <td>-1.841663e+00</td>\n",
       "      <td>-1.652227e+00</td>\n",
       "      <td>-1.810713e+00</td>\n",
       "      <td>-1.875352e+00</td>\n",
       "      <td>-1.515879e+00</td>\n",
       "      <td>-1.866748e+00</td>\n",
       "      <td>-2.355239e+01</td>\n",
       "      <td>-1.875925e+01</td>\n",
       "      <td>-2.196547e+01</td>\n",
       "      <td>-1.969503e+01</td>\n",
       "      <td>-2.065124e+01</td>\n",
       "      <td>-2.343220e+01</td>\n",
       "      <td>-3.843378e+01</td>\n",
       "      <td>-2.147554e+01</td>\n",
       "      <td>-2.268682e+01</td>\n",
       "      <td>-2.967671e+01</td>\n",
       "      <td>-2.335012e+01</td>\n",
       "      <td>-1.652946e+01</td>\n",
       "      <td>-2.020461e+01</td>\n",
       "      <td>-2.255056e+01</td>\n",
       "      <td>-3.686078e+01</td>\n",
       "      <td>-3.847736e+01</td>\n",
       "      <td>...</td>\n",
       "      <td>-2.433150e+01</td>\n",
       "      <td>-2.362536e+01</td>\n",
       "      <td>-4.573577e+01</td>\n",
       "      <td>-2.637486e+01</td>\n",
       "      <td>-2.917145e+01</td>\n",
       "      <td>-1.682569e+01</td>\n",
       "      <td>-1.682179e+01</td>\n",
       "      <td>-1.766429e+01</td>\n",
       "      <td>-1.810442e+01</td>\n",
       "      <td>-1.715994e+01</td>\n",
       "      <td>-3.210613e+01</td>\n",
       "      <td>-2.300836e+01</td>\n",
       "      <td>-1.548023e+01</td>\n",
       "      <td>-3.319885e+01</td>\n",
       "      <td>-3.319514e+01</td>\n",
       "      <td>-3.038489e+01</td>\n",
       "      <td>-1.204649e+01</td>\n",
       "      <td>-2.226450e+01</td>\n",
       "      <td>-1.829168e+01</td>\n",
       "      <td>-2.529849e+01</td>\n",
       "      <td>-2.803944e+01</td>\n",
       "      <td>-3.034925e+01</td>\n",
       "      <td>-5.106798e+01</td>\n",
       "      <td>-2.849609e+01</td>\n",
       "      <td>-2.111252e+01</td>\n",
       "      <td>-2.706919e+01</td>\n",
       "      <td>-3.659346e+01</td>\n",
       "      <td>-1.422095e+01</td>\n",
       "      <td>-4.128419e+01</td>\n",
       "      <td>-3.856717e+01</td>\n",
       "      <td>-1.413267e+01</td>\n",
       "      <td>-2.468577e+01</td>\n",
       "      <td>-1.520687e+01</td>\n",
       "      <td>-3.305562e+01</td>\n",
       "      <td>-2.117983e+01</td>\n",
       "      <td>-2.786972e+01</td>\n",
       "      <td>-2.464943e+01</td>\n",
       "      <td>-1.805304e+01</td>\n",
       "      <td>-4.045652e+01</td>\n",
       "      <td>-1.449495e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-5.638876e-01</td>\n",
       "      <td>-5.311584e-01</td>\n",
       "      <td>-5.697513e-01</td>\n",
       "      <td>-5.905186e-01</td>\n",
       "      <td>-6.179717e-01</td>\n",
       "      <td>-6.935698e-01</td>\n",
       "      <td>-5.755554e-01</td>\n",
       "      <td>-5.862542e-01</td>\n",
       "      <td>-5.688679e-01</td>\n",
       "      <td>-6.624309e-01</td>\n",
       "      <td>-5.732855e-01</td>\n",
       "      <td>-6.131185e-01</td>\n",
       "      <td>-6.951746e-01</td>\n",
       "      <td>-6.419074e-01</td>\n",
       "      <td>-6.759057e-01</td>\n",
       "      <td>-6.390212e-01</td>\n",
       "      <td>-6.390820e-01</td>\n",
       "      <td>-6.777498e-01</td>\n",
       "      <td>-6.664736e-01</td>\n",
       "      <td>-6.420709e-01</td>\n",
       "      <td>-6.365427e-01</td>\n",
       "      <td>-6.744530e-01</td>\n",
       "      <td>-6.491041e-01</td>\n",
       "      <td>-6.149120e-01</td>\n",
       "      <td>-4.493756e-01</td>\n",
       "      <td>-4.635728e-01</td>\n",
       "      <td>-3.744862e-01</td>\n",
       "      <td>-4.156852e-01</td>\n",
       "      <td>-4.593146e-01</td>\n",
       "      <td>-5.116203e-01</td>\n",
       "      <td>-4.403281e-01</td>\n",
       "      <td>-4.248182e-01</td>\n",
       "      <td>-4.611867e-01</td>\n",
       "      <td>-4.578923e-01</td>\n",
       "      <td>-4.124200e-01</td>\n",
       "      <td>-5.390717e-01</td>\n",
       "      <td>-4.136821e-01</td>\n",
       "      <td>-4.785796e-01</td>\n",
       "      <td>-3.822540e-01</td>\n",
       "      <td>-4.504104e-01</td>\n",
       "      <td>...</td>\n",
       "      <td>-4.819230e-01</td>\n",
       "      <td>-4.976248e-01</td>\n",
       "      <td>-4.397595e-01</td>\n",
       "      <td>-4.269083e-01</td>\n",
       "      <td>-3.237309e-01</td>\n",
       "      <td>-4.841597e-01</td>\n",
       "      <td>-4.615884e-01</td>\n",
       "      <td>-4.371545e-01</td>\n",
       "      <td>-4.929491e-01</td>\n",
       "      <td>-4.600017e-01</td>\n",
       "      <td>-4.795339e-01</td>\n",
       "      <td>-4.462117e-01</td>\n",
       "      <td>-4.839344e-01</td>\n",
       "      <td>-4.044806e-01</td>\n",
       "      <td>-3.880277e-01</td>\n",
       "      <td>-4.075018e-01</td>\n",
       "      <td>-4.957911e-01</td>\n",
       "      <td>-4.042848e-01</td>\n",
       "      <td>-4.855927e-01</td>\n",
       "      <td>-4.151934e-01</td>\n",
       "      <td>-3.462071e-01</td>\n",
       "      <td>-4.163205e-01</td>\n",
       "      <td>-3.595521e-01</td>\n",
       "      <td>-4.896345e-01</td>\n",
       "      <td>-4.511315e-01</td>\n",
       "      <td>-3.588562e-01</td>\n",
       "      <td>-4.708746e-01</td>\n",
       "      <td>-4.344471e-01</td>\n",
       "      <td>-2.916292e-01</td>\n",
       "      <td>-4.425110e-01</td>\n",
       "      <td>-5.466954e-01</td>\n",
       "      <td>-3.763002e-01</td>\n",
       "      <td>-5.118673e-01</td>\n",
       "      <td>-4.478800e-01</td>\n",
       "      <td>-4.397373e-01</td>\n",
       "      <td>-4.299846e-01</td>\n",
       "      <td>-4.307504e-01</td>\n",
       "      <td>-5.268242e-01</td>\n",
       "      <td>-4.292140e-01</td>\n",
       "      <td>-4.598088e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.442775e-01</td>\n",
       "      <td>1.377638e-01</td>\n",
       "      <td>5.161888e-02</td>\n",
       "      <td>-1.115648e-01</td>\n",
       "      <td>2.275066e-02</td>\n",
       "      <td>-1.280171e-01</td>\n",
       "      <td>2.309700e-02</td>\n",
       "      <td>7.320083e-03</td>\n",
       "      <td>9.665689e-03</td>\n",
       "      <td>-1.489907e-02</td>\n",
       "      <td>1.064278e-02</td>\n",
       "      <td>-2.832901e-02</td>\n",
       "      <td>-2.070092e-01</td>\n",
       "      <td>-2.455417e-01</td>\n",
       "      <td>-2.220075e-01</td>\n",
       "      <td>-2.499821e-01</td>\n",
       "      <td>-1.967996e-01</td>\n",
       "      <td>-2.416020e-01</td>\n",
       "      <td>-1.959821e-01</td>\n",
       "      <td>-2.336870e-01</td>\n",
       "      <td>-2.061242e-01</td>\n",
       "      <td>-1.984461e-01</td>\n",
       "      <td>-2.502268e-01</td>\n",
       "      <td>-1.817949e-01</td>\n",
       "      <td>-1.266212e-01</td>\n",
       "      <td>-1.206790e-01</td>\n",
       "      <td>1.372887e-02</td>\n",
       "      <td>1.965026e-02</td>\n",
       "      <td>-4.398765e-02</td>\n",
       "      <td>-1.080744e-01</td>\n",
       "      <td>-1.039974e-02</td>\n",
       "      <td>-1.267623e-02</td>\n",
       "      <td>-2.402420e-02</td>\n",
       "      <td>-1.039665e-01</td>\n",
       "      <td>6.226814e-02</td>\n",
       "      <td>-1.525126e-01</td>\n",
       "      <td>-5.164891e-02</td>\n",
       "      <td>-8.316110e-02</td>\n",
       "      <td>1.128178e-01</td>\n",
       "      <td>-9.200374e-03</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.347085e-02</td>\n",
       "      <td>-3.418526e-02</td>\n",
       "      <td>-5.398361e-02</td>\n",
       "      <td>-1.315678e-02</td>\n",
       "      <td>1.424748e-01</td>\n",
       "      <td>-1.248806e-01</td>\n",
       "      <td>-6.082867e-02</td>\n",
       "      <td>-6.022220e-02</td>\n",
       "      <td>-6.531348e-02</td>\n",
       "      <td>-2.276454e-02</td>\n",
       "      <td>-5.982990e-02</td>\n",
       "      <td>2.751061e-02</td>\n",
       "      <td>-7.488806e-02</td>\n",
       "      <td>7.243625e-02</td>\n",
       "      <td>4.084039e-02</td>\n",
       "      <td>3.523497e-02</td>\n",
       "      <td>-8.535268e-03</td>\n",
       "      <td>-8.404490e-02</td>\n",
       "      <td>-9.020841e-02</td>\n",
       "      <td>1.888310e-03</td>\n",
       "      <td>9.409576e-02</td>\n",
       "      <td>3.146517e-02</td>\n",
       "      <td>4.577475e-02</td>\n",
       "      <td>-7.307116e-03</td>\n",
       "      <td>7.474668e-02</td>\n",
       "      <td>1.185304e-01</td>\n",
       "      <td>-1.230538e-01</td>\n",
       "      <td>5.269330e-02</td>\n",
       "      <td>1.667031e-01</td>\n",
       "      <td>-1.049565e-03</td>\n",
       "      <td>-2.051342e-01</td>\n",
       "      <td>1.151222e-01</td>\n",
       "      <td>-1.043898e-01</td>\n",
       "      <td>-4.517888e-02</td>\n",
       "      <td>2.990574e-02</td>\n",
       "      <td>-1.788398e-02</td>\n",
       "      <td>3.008970e-02</td>\n",
       "      <td>-1.001326e-01</td>\n",
       "      <td>-6.574410e-02</td>\n",
       "      <td>-5.763047e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>7.317100e-01</td>\n",
       "      <td>6.758454e-01</td>\n",
       "      <td>5.980758e-01</td>\n",
       "      <td>4.668744e-01</td>\n",
       "      <td>6.240130e-01</td>\n",
       "      <td>5.564446e-01</td>\n",
       "      <td>6.104375e-01</td>\n",
       "      <td>5.901695e-01</td>\n",
       "      <td>5.894924e-01</td>\n",
       "      <td>6.525343e-01</td>\n",
       "      <td>5.907976e-01</td>\n",
       "      <td>5.793166e-01</td>\n",
       "      <td>4.358331e-01</td>\n",
       "      <td>3.439870e-01</td>\n",
       "      <td>4.123615e-01</td>\n",
       "      <td>3.435088e-01</td>\n",
       "      <td>3.989686e-01</td>\n",
       "      <td>4.066661e-01</td>\n",
       "      <td>4.317881e-01</td>\n",
       "      <td>3.536420e-01</td>\n",
       "      <td>3.836953e-01</td>\n",
       "      <td>4.342104e-01</td>\n",
       "      <td>3.615573e-01</td>\n",
       "      <td>3.741295e-01</td>\n",
       "      <td>3.758779e-01</td>\n",
       "      <td>3.356775e-01</td>\n",
       "      <td>3.924052e-01</td>\n",
       "      <td>4.288088e-01</td>\n",
       "      <td>4.257723e-01</td>\n",
       "      <td>3.909540e-01</td>\n",
       "      <td>4.383029e-01</td>\n",
       "      <td>4.399820e-01</td>\n",
       "      <td>4.221423e-01</td>\n",
       "      <td>3.074795e-01</td>\n",
       "      <td>4.768112e-01</td>\n",
       "      <td>4.088835e-01</td>\n",
       "      <td>3.289584e-01</td>\n",
       "      <td>4.179632e-01</td>\n",
       "      <td>4.965377e-01</td>\n",
       "      <td>4.477446e-01</td>\n",
       "      <td>...</td>\n",
       "      <td>4.722113e-01</td>\n",
       "      <td>4.537573e-01</td>\n",
       "      <td>3.802789e-01</td>\n",
       "      <td>4.260445e-01</td>\n",
       "      <td>4.979216e-01</td>\n",
       "      <td>3.629212e-01</td>\n",
       "      <td>3.882181e-01</td>\n",
       "      <td>3.226990e-01</td>\n",
       "      <td>4.285659e-01</td>\n",
       "      <td>4.066440e-01</td>\n",
       "      <td>4.031240e-01</td>\n",
       "      <td>4.755073e-01</td>\n",
       "      <td>4.060183e-01</td>\n",
       "      <td>4.530141e-01</td>\n",
       "      <td>4.780044e-01</td>\n",
       "      <td>4.192291e-01</td>\n",
       "      <td>4.857568e-01</td>\n",
       "      <td>2.854014e-01</td>\n",
       "      <td>3.849043e-01</td>\n",
       "      <td>4.246213e-01</td>\n",
       "      <td>4.712277e-01</td>\n",
       "      <td>4.293974e-01</td>\n",
       "      <td>4.251828e-01</td>\n",
       "      <td>4.622619e-01</td>\n",
       "      <td>5.105105e-01</td>\n",
       "      <td>4.982904e-01</td>\n",
       "      <td>3.183511e-01</td>\n",
       "      <td>5.150252e-01</td>\n",
       "      <td>4.859609e-01</td>\n",
       "      <td>4.736714e-01</td>\n",
       "      <td>3.273959e-01</td>\n",
       "      <td>4.942984e-01</td>\n",
       "      <td>3.940738e-01</td>\n",
       "      <td>4.204911e-01</td>\n",
       "      <td>5.045541e-01</td>\n",
       "      <td>4.385694e-01</td>\n",
       "      <td>4.516087e-01</td>\n",
       "      <td>4.131805e-01</td>\n",
       "      <td>3.595684e-01</td>\n",
       "      <td>3.786469e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>3.057122e+00</td>\n",
       "      <td>7.416766e+00</td>\n",
       "      <td>8.909792e+00</td>\n",
       "      <td>1.765551e+01</td>\n",
       "      <td>1.175114e+01</td>\n",
       "      <td>1.007897e+01</td>\n",
       "      <td>1.198797e+01</td>\n",
       "      <td>1.343956e+01</td>\n",
       "      <td>1.347767e+01</td>\n",
       "      <td>8.964296e+00</td>\n",
       "      <td>7.090614e+00</td>\n",
       "      <td>1.025687e+01</td>\n",
       "      <td>2.321918e+01</td>\n",
       "      <td>3.621242e+01</td>\n",
       "      <td>2.762232e+01</td>\n",
       "      <td>2.775638e+01</td>\n",
       "      <td>3.985768e+01</td>\n",
       "      <td>2.766484e+01</td>\n",
       "      <td>3.177211e+01</td>\n",
       "      <td>2.933540e+01</td>\n",
       "      <td>4.308266e+01</td>\n",
       "      <td>2.056735e+01</td>\n",
       "      <td>3.465084e+01</td>\n",
       "      <td>6.193636e+01</td>\n",
       "      <td>1.649785e+01</td>\n",
       "      <td>3.416372e+01</td>\n",
       "      <td>2.670132e+01</td>\n",
       "      <td>1.576088e+01</td>\n",
       "      <td>1.998443e+01</td>\n",
       "      <td>2.601625e+01</td>\n",
       "      <td>1.838174e+01</td>\n",
       "      <td>2.691332e+01</td>\n",
       "      <td>3.865847e+01</td>\n",
       "      <td>4.366883e+01</td>\n",
       "      <td>4.273974e+01</td>\n",
       "      <td>2.245118e+01</td>\n",
       "      <td>3.417137e+01</td>\n",
       "      <td>1.753563e+01</td>\n",
       "      <td>2.398130e+01</td>\n",
       "      <td>4.057395e+01</td>\n",
       "      <td>...</td>\n",
       "      <td>1.517978e+01</td>\n",
       "      <td>3.237494e+01</td>\n",
       "      <td>3.403535e+01</td>\n",
       "      <td>3.472804e+01</td>\n",
       "      <td>1.003386e+01</td>\n",
       "      <td>2.520328e+01</td>\n",
       "      <td>2.833136e+01</td>\n",
       "      <td>4.855057e+01</td>\n",
       "      <td>3.186487e+01</td>\n",
       "      <td>3.969876e+01</td>\n",
       "      <td>3.049951e+01</td>\n",
       "      <td>1.980749e+01</td>\n",
       "      <td>2.159619e+01</td>\n",
       "      <td>3.638141e+01</td>\n",
       "      <td>1.295372e+01</td>\n",
       "      <td>2.687900e+01</td>\n",
       "      <td>2.056303e+01</td>\n",
       "      <td>4.080572e+01</td>\n",
       "      <td>4.962670e+01</td>\n",
       "      <td>2.031260e+01</td>\n",
       "      <td>1.577953e+01</td>\n",
       "      <td>2.065479e+01</td>\n",
       "      <td>1.312081e+01</td>\n",
       "      <td>2.613535e+01</td>\n",
       "      <td>2.009278e+01</td>\n",
       "      <td>2.834790e+01</td>\n",
       "      <td>3.747219e+01</td>\n",
       "      <td>2.352090e+01</td>\n",
       "      <td>2.405977e+01</td>\n",
       "      <td>2.990595e+01</td>\n",
       "      <td>2.571906e+01</td>\n",
       "      <td>2.590152e+01</td>\n",
       "      <td>2.602152e+01</td>\n",
       "      <td>1.723191e+01</td>\n",
       "      <td>1.610766e+01</td>\n",
       "      <td>3.187494e+01</td>\n",
       "      <td>1.641892e+01</td>\n",
       "      <td>3.445501e+01</td>\n",
       "      <td>3.989141e+01</td>\n",
       "      <td>2.719132e+01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         TimbreAvg1         ...          TimbreCovariance78\n",
       "count  4.637150e+05         ...                4.637150e+05\n",
       "mean   7.860182e-16         ...                4.832588e-15\n",
       "std    1.000001e+00         ...                1.000001e+00\n",
       "min   -6.849083e+00         ...               -1.449495e+01\n",
       "25%   -5.638876e-01         ...               -4.598088e-01\n",
       "50%    1.442775e-01         ...               -5.763047e-02\n",
       "75%    7.317100e-01         ...                3.786469e-01\n",
       "max    3.057122e+00         ...                2.719132e+01\n",
       "\n",
       "[8 rows x 90 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_std = pd.DataFrame(X_train_std,columns=X_train.columns)\n",
    "X_train_std.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our values are normalised and so we now actually apply PCA\n",
    "(https://towardsdatascience.com/pca-using-python-scikit-learn-e653f8989e60)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "55"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "# Make an instance of the Model\n",
    "pca = PCA(.90)\n",
    "\n",
    "# We fit to only our training set\n",
    "pca.fit(X_train_std)\n",
    "# Print number of components generated\n",
    "pca.n_components_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "PCA(.90) means that scikit-learn choose the minimum number of principal components such that 90% of the variance is retained.\n",
    "\n",
    "In this case, 90% of the variance amounts to 55 principal components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_proc = pca.transform(X_train_std)\n",
    "X_test_proc = pca.transform(X_test_std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_proc = y_train - min(y_train)\n",
    "y_test_proc = y_test - min(y_test)\n",
    "# y_train_proc"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.python.keras import Sequential\n",
    "from tensorflow.python.keras.layers import Dense, Lambda, Dropout\n",
    "# from tensorflow.python.keras.initializers import Initializer\n",
    "from tensorflow.python.keras.utils import to_categorical\n",
    "from tensorflow.python.keras.callbacks import ReduceLROnPlateau, EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_hot = to_categorical(y_train_proc, 90)\n",
    "y_test_hot = to_categorical(y_test_proc, 90)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(463715, 55)\n",
      "(51630, 90)\n"
     ]
    }
   ],
   "source": [
    "print(X_train_proc.shape)\n",
    "print(y_test_hot.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot(history):\n",
    "    epochs = range(1, len(history.history['loss']) + 1)\n",
    "    plt.plot(epochs, history.history['mean_absolute_error'], label='train');\n",
    "    plt.plot(epochs, history.history['val_mean_absolute_error'], label='val');\n",
    "    plt.xlabel('epoch');\n",
    "    plt.ylabel('mae');\n",
    "    plt.legend();\n",
    "    plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/ops/resource_variable_ops.py:435: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "model1 = Sequential()\n",
    "model1.add(Dense(55, input_shape=(55,)))\n",
    "model1.add(Dense(110))\n",
    "model1.add(Dense(90, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "learning_rate_reduction = ReduceLROnPlateau(monitor='val_acc', \n",
    "                                            patience=4, \n",
    "                                            verbose=1, \n",
    "                                            factor=0.5, \n",
    "                                            min_lr=0.0001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "model1.compile(optimizer='adam'\n",
    "             , loss='categorical_crossentropy'\n",
    "             , metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 463715 samples, validate on 51630 samples\n",
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Epoch 1/5\n",
      "463715/463715 [==============================] - 22s 47us/sample - loss: 3.3497 - acc: 0.0754 - val_loss: 3.8857 - val_acc: 0.0060\n",
      "Epoch 2/5\n",
      "463715/463715 [==============================] - 23s 50us/sample - loss: 3.3127 - acc: 0.0766 - val_loss: 3.8507 - val_acc: 0.0104\n",
      "Epoch 3/5\n",
      "463715/463715 [==============================] - 23s 50us/sample - loss: 3.3094 - acc: 0.0775 - val_loss: 3.8464 - val_acc: 0.0038\n",
      "Epoch 4/5\n",
      "463715/463715 [==============================] - 21s 46us/sample - loss: 3.3078 - acc: 0.0777 - val_loss: 3.8373 - val_acc: 0.0063\n",
      "Epoch 5/5\n",
      "463715/463715 [==============================] - 21s 46us/sample - loss: 3.3064 - acc: 0.0775 - val_loss: 3.8320 - val_acc: 0.0077\n"
     ]
    }
   ],
   "source": [
    "fit1 = model1.fit(x=X_train_proc, y=y_train_hot\n",
    "          , epochs=5\n",
    "          , batch_size=64\n",
    "          , validation_data=(X_test_proc, y_test_hot)\n",
    "          , callbacks=[learning_rate_reduction])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plot(fit1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 55)                3080      \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 110)               6160      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 90)                9990      \n",
      "=================================================================\n",
      "Total params: 19,230\n",
      "Trainable params: 19,230\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds = model1.predict_classes(X_test_proc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[80 76 78 ... 79 79 78]\n",
      "[83 85 85 ... 85 85 85]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "12.18074762734844"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(np.array(y_test_proc))\n",
    "print(preds)\n",
    "np.mean(np.absolute((preds-np.array(y_test_proc))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "model2 = Sequential()\n",
    "model2.add(Dense(55, input_shape=(55,), activation='relu'))\n",
    "model2.add(Dense(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "learning_rate_reduction1 = ReduceLROnPlateau(monitor='mean_absolute_error', \n",
    "                                            patience=2, \n",
    "                                            verbose=1, \n",
    "                                            factor=0.5, \n",
    "                                            min_lr=0.0001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/keras/utils/losses_utils.py:170: to_float (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n"
     ]
    }
   ],
   "source": [
    "model2.compile(optimizer='adam'\n",
    "             , loss='mse'\n",
    "             , metrics=['mae'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "_kg_hide-output": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 463715 samples, validate on 51630 samples\n",
      "Epoch 1/10\n",
      "463715/463715 [==============================] - 19s 40us/sample - loss: 454.0315 - mean_absolute_error: 13.8813 - val_loss: 126.2524 - val_mean_absolute_error: 7.8502\n",
      "Epoch 2/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 96.3439 - mean_absolute_error: 7.1983 - val_loss: 119.6077 - val_mean_absolute_error: 7.5977\n",
      "Epoch 3/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 92.9060 - mean_absolute_error: 7.0213 - val_loss: 121.2777 - val_mean_absolute_error: 7.7042\n",
      "Epoch 4/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 91.5295 - mean_absolute_error: 6.9511 - val_loss: 116.8438 - val_mean_absolute_error: 7.4562\n",
      "Epoch 5/10\n",
      "463715/463715 [==============================] - 18s 39us/sample - loss: 90.6753 - mean_absolute_error: 6.9032 - val_loss: 118.6354 - val_mean_absolute_error: 7.6665\n",
      "Epoch 6/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 90.0552 - mean_absolute_error: 6.8741 - val_loss: 120.3845 - val_mean_absolute_error: 7.6576\n",
      "Epoch 7/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 89.7168 - mean_absolute_error: 6.8559 - val_loss: 118.5803 - val_mean_absolute_error: 7.6459\n",
      "Epoch 8/10\n",
      "463715/463715 [==============================] - 17s 37us/sample - loss: 89.4038 - mean_absolute_error: 6.8436 - val_loss: 115.2567 - val_mean_absolute_error: 7.5358\n",
      "Epoch 9/10\n",
      "463715/463715 [==============================] - 18s 38us/sample - loss: 89.1831 - mean_absolute_error: 6.8339 - val_loss: 114.6454 - val_mean_absolute_error: 7.4279\n",
      "Epoch 10/10\n",
      "463715/463715 [==============================] - 18s 40us/sample - loss: 88.9885 - mean_absolute_error: 6.8232 - val_loss: 115.5709 - val_mean_absolute_error: 7.5337\n"
     ]
    }
   ],
   "source": [
    "fit2 = model2.fit(x=X_train_proc, y=y_train_proc\n",
    "          , epochs=10\n",
    "          , batch_size=64\n",
    "          , validation_data=(X_test_proc, y_test_proc)\n",
    "          , callbacks=[learning_rate_reduction1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.533661957422523"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds_model_rms = model2.predict(X_test_proc)\n",
    "np.mean(np.absolute(preds_model_rms.T-np.array(y_test_proc)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot(fit2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from sklearn.metrics import mean_squared_error, r2_score\n",
    "# mean_squared_error(predictions_linearRegr, np.array(y_test_proc))\n",
    "es = EarlyStopping(monitor='val_mean_absolute_error', patience=2, restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /opt/conda/lib/python3.6/site-packages/tensorflow/python/keras/layers/core.py:143: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n"
     ]
    }
   ],
   "source": [
    "model3 = Sequential()\n",
    "model3.add(Dense(55, input_shape=(55,), activation='relu'))\n",
    "model3.add(Dense(110, activation='relu'))\n",
    "model3.add(Dropout(0.2))\n",
    "model3.add(Dense(1))\n",
    "\n",
    "model3.compile(optimizer='adam'\n",
    "             , loss='mse'\n",
    "             , metrics=['mae'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 463715 samples, validate on 51630 samples\n",
      "Epoch 1/10\n",
      "463715/463715 [==============================] - 20s 44us/sample - loss: 220.8219 - mean_absolute_error: 10.4302 - val_loss: 119.6071 - val_mean_absolute_error: 7.7226\n",
      "Epoch 2/10\n",
      "463715/463715 [==============================] - 20s 44us/sample - loss: 117.5109 - mean_absolute_error: 8.2226 - val_loss: 122.8387 - val_mean_absolute_error: 7.8060\n",
      "Epoch 3/10\n",
      "463715/463715 [==============================] - 20s 43us/sample - loss: 114.2269 - mean_absolute_error: 8.0849 - val_loss: 105.3596 - val_mean_absolute_error: 7.0479\n",
      "Epoch 4/10\n",
      "463715/463715 [==============================] - 23s 50us/sample - loss: 111.7011 - mean_absolute_error: 7.9776 - val_loss: 117.5443 - val_mean_absolute_error: 7.5872\n",
      "Epoch 5/10\n",
      "463715/463715 [==============================] - 20s 43us/sample - loss: 110.0584 - mean_absolute_error: 7.9063 - val_loss: 111.5645 - val_mean_absolute_error: 7.4130\n"
     ]
    }
   ],
   "source": [
    "fit3 = model3.fit(x=X_train_proc, y=y_train_proc\n",
    "          , epochs=10\n",
    "          , batch_size=64\n",
    "          , validation_data=(X_test_proc, y_test_proc)\n",
    "          , callbacks=[learning_rate_reduction1, es])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.047857569004246"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds_model_rms = model3.predict(X_test_proc)\n",
    "np.mean(np.absolute(preds_model_rms.T-np.array(y_test_proc)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot(fit3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "batch size = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 463715 samples, validate on 51630 samples\n",
      "Epoch 1/10\n",
      "463715/463715 [==============================] - 11s 23us/sample - loss: 111.0970 - mean_absolute_error: 7.9487 - val_loss: 109.0630 - val_mean_absolute_error: 7.2164\n",
      "Epoch 2/10\n",
      "463715/463715 [==============================] - 10s 22us/sample - loss: 110.0095 - mean_absolute_error: 7.9035 - val_loss: 111.0600 - val_mean_absolute_error: 7.3114\n",
      "Epoch 3/10\n",
      "463715/463715 [==============================] - 10s 22us/sample - loss: 108.6083 - mean_absolute_error: 7.8469 - val_loss: 112.4737 - val_mean_absolute_error: 7.3348\n"
     ]
    }
   ],
   "source": [
    "model3.compile(optimizer='adam'\n",
    "             , loss='mse'\n",
    "             , metrics=['mae'])\n",
    "fit3 = model3.fit(x=X_train_proc, y=y_train_proc\n",
    "          , epochs=10\n",
    "          , batch_size=128\n",
    "          , validation_data=(X_test_proc, y_test_proc)\n",
    "          , callbacks=[learning_rate_reduction1, es])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.216405226910458"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds_model_rms = model3.predict(X_test_proc)\n",
    "np.mean(np.absolute(preds_model_rms.T-np.array(y_test_proc)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot(fit3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "batch size = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.optimizers import RMSprop\n",
    "# adam = optimizers.Adam()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 463715 samples, validate on 51630 samples\n",
      "Epoch 1/10\n",
      "463715/463715 [==============================] - 20s 43us/sample - loss: 110.6858 - mean_absolute_error: 7.9279 - val_loss: 133.4714 - val_mean_absolute_error: 8.3044\n",
      "Epoch 2/10\n",
      "463715/463715 [==============================] - 18s 39us/sample - loss: 109.3629 - mean_absolute_error: 7.8753 - val_loss: 116.5361 - val_mean_absolute_error: 7.5751\n",
      "Epoch 3/10\n",
      "463715/463715 [==============================] - 18s 39us/sample - loss: 108.0603 - mean_absolute_error: 7.8145 - val_loss: 99.5788 - val_mean_absolute_error: 6.8666\n",
      "Epoch 4/10\n",
      "463715/463715 [==============================] - 18s 39us/sample - loss: 106.9977 - mean_absolute_error: 7.7635 - val_loss: 114.6966 - val_mean_absolute_error: 7.4764\n",
      "Epoch 5/10\n",
      "463715/463715 [==============================] - 20s 42us/sample - loss: 105.5895 - mean_absolute_error: 7.7047 - val_loss: 122.0308 - val_mean_absolute_error: 7.7521\n"
     ]
    }
   ],
   "source": [
    "model3.compile(optimizer='RMSprop'\n",
    "             , loss='mse'\n",
    "             , metrics=['mae'])\n",
    "fit4 = model3.fit(x=X_train_proc, y=y_train_proc\n",
    "          , epochs=10\n",
    "          , batch_size=64\n",
    "          , validation_data=(X_test_proc, y_test_proc)\n",
    "          , callbacks=[learning_rate_reduction1, es])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6.866590757072607"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds_model_rms = model3.predict(X_test_proc)\n",
    "np.mean(np.absolute(preds_model_rms.T-np.array(y_test_proc)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "cm = confusion_matrix(np.round_(np.array(preds_model_rms), decimals=-1), np.round_(np.array(y_test_proc), decimals=-1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>1920</th>\n",
       "      <th>1930</th>\n",
       "      <th>1940</th>\n",
       "      <th>1950</th>\n",
       "      <th>1960</th>\n",
       "      <th>1970</th>\n",
       "      <th>1980</th>\n",
       "      <th>1990</th>\n",
       "      <th>2000</th>\n",
       "      <th>2010</th>\n",
       "      <th>2020</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1920</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1930</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1940</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1950</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1960</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1970</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>41</td>\n",
       "      <td>25</td>\n",
       "      <td>17</td>\n",
       "      <td>30</td>\n",
       "      <td>27</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1980</th>\n",
       "      <td>22</td>\n",
       "      <td>33</td>\n",
       "      <td>21</td>\n",
       "      <td>120</td>\n",
       "      <td>532</td>\n",
       "      <td>508</td>\n",
       "      <td>848</td>\n",
       "      <td>928</td>\n",
       "      <td>783</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1990</th>\n",
       "      <td>1</td>\n",
       "      <td>12</td>\n",
       "      <td>50</td>\n",
       "      <td>160</td>\n",
       "      <td>1021</td>\n",
       "      <td>1433</td>\n",
       "      <td>3923</td>\n",
       "      <td>6059</td>\n",
       "      <td>6644</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>196</td>\n",
       "      <td>410</td>\n",
       "      <td>1204</td>\n",
       "      <td>7187</td>\n",
       "      <td>19303</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2010</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>8</td>\n",
       "      <td>53</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2020</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      1920  1930  1940  1950  1960  1970  1980  1990   2000  2010  2020\n",
       "1920     0     0     0     0     0     0     0     0      0     0     0\n",
       "1930     0     0     0     0     0     0     0     0      0     0     0\n",
       "1940     0     0     0     0     0     0     0     0      0     0     0\n",
       "1950     0     0     0     0     0     0     0     0      0     0     0\n",
       "1960     0     0     0     0     0     0     0     0      0     0     0\n",
       "1970     3     1     0     1    41    25    17    30     27     0     0\n",
       "1980    22    33    21   120   532   508   848   928    783     0     0\n",
       "1990     1    12    50   160  1021  1433  3923  6059   6644     0     0\n",
       "2000     0     0     0    21   196   410  1204  7187  19303     0     0\n",
       "2010     0     0     0     0     0     1     2     8     53     0     0\n",
       "2020     0     0     0     0     0     0     0     0      2     0     0"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAELCAYAAADawD2zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzt3Xu4XFWd5vHvm4RAINxEQQ1g4hB0wkWUAHkUFUFuahNs0QYdiJoBbXDQxgsJdre3dgamaVEfAc0DgeAtMgw0aQUCraijmBAgJJBwMYBC0kgEAjQgCefUb/7Yq0xxPCdn16m969Suej959pOqtXettXfVOavWWXut9VNEYGZm3W/MaJ+AmZm1hyt8M7Me4QrfzKxHuMI3M+sRrvDNzHqEK3wzsx7hCt/MrEe4wjcz6xGu8M3MesS40T6B4YwbP8lTgc1sWH2b1qnVPF58/MHc9c1WL39ty+W1W8dX+GZmbVPrH+0zKJUrfDOzuqiN9hmUyhW+mVldzRW+mVlPCLfwzcx6hFv4ZmY9ov/F0T6DUg07Dl/SfEnrJd3dkPYGSb+RdJekf5O0Q0o/UtLtKf12SYc3vObAlL5G0jclVW5Ik5l1uajl3yooz8Sry4FjBqRdAsyJiP2Aa4DPpvTHgb9K6bOA7za85mLgVGBq2gbmaWY2umq1/FsFDVvhR8QvgScHJO8N/DI9vgl4Xzp2eUT8R0pfBUyQtLWkVwE7RMSSyGIqXgEcX8QFmJkVJaKWe6uikS6tsAqYmR6/H9hjkGPeB9wRERuBScDahn1rU9qgJJ0m6TZJt9Vqz43wFM3MmtTrLfwhfBQ4XdLtwPbApsadkvYBzgM+NpLMI2JeREyPiOljxmw3wlM0M2tSl/fhj2iUTkTcCxwFIGlv4N31fZJ2J+vXPyUiHkjJ64DdG7LYPaWZmXWOXh+lMxhJu6b/xwB/D3w7Pd8J+AnZDd1f14+PiEeBZyTNSKNzTgGubfHczcyK1etdOpJ+CPwGeJ2ktZJmAydJuh+4F/gP4LJ0+CeAvYB/lHRn2nZN+04nG92zBngAuL7YSzEza1GXd+koGzTTubw8spnlUcTyyBtXLs5d32y9/9GVm0vkmbZmZkmEl0c2M+sN/X2jfQalcoVvZlZX0b75vFzhm5nVOeKVmVmPcAvfzKxHVHR8fV6u8M3M6tzCNzPrEX0epWNm1hM8Dt/MrFd0eR9+0SEOD25YQ2eFpPc2vOYYSfelEIdzyrkcM7MWdPlaOkWHOLwbmB4RB6TXfEfSOEljgQuBY4FpZIuvTSvg/M3MitPrq2U2GeLw+Yio3/XYBqgvRHQwsCYiHoyITcBCNkfMMjPrDG7hD2rIEIeSDpG0CrgL+Hj6ApgEPNLw+i2GODQzGxX9ffm3Cio8xGFELI2IfYCDgLmStmk2c8e0NbNR0eVdOoWHOGw45h5JzwL7koUzbAx0vsUQhxExD5gHXg/fzNqoohV5XkWHOJwiaVx6/Brg9cDvgGXA1LR/PHAisKjlszczK1KX9+EP28JPIQ4PA14uaS3wBWCipDPSIVezOcThocAcSS8CNeD0iHg85fMJYDEwFpgfEauKvBAzs5Z1eQvfIQ7NrCsUEeLwT9ecm7u+mfDeOVssT9J84D3A+ojYtyH9fwBnAP3ATyLicyl9LjA7pZ8ZEYtT+jHAN8gay5dExLkpfQrZiMddgNuBk9MoyCGN9KatmVn3KbZL53IGzGGS9A6yEY5vSINbzk/p08i6uvdJr7lI0thh5jCdB1wQEXsBG8i+LLbIFb6ZWV2Bo3SGmMP0t8C5EbExHbM+pc8EFkbExoh4CFhDNn9p0DlMkgQcDlyVXr8AOH64c3KFb2ZW10SF3zh8PG2n5Shhb+CtkpZK+oWkg1L6UHOVhkrfBXiqYaJrrrlNXjzNzKyuiXuajcPHmzAOeBkwg2yu0pWSXttkHiPmCt/MrK78UTprgasjGy1zq6Qa8HK2PFdpsPQngJ0kjUut/C3Obapzl46ZWV35Syv8K/AO+POk1fHA42Tzkk6UtHUafTMVuJUh5jClL4ybgRNSvrOAa4cr3C18M7O6Alv4Q8xhmg/MT8vNbwJmpcp7laQrgdVAH3BGpGgsW5jDdDawUNI/AcuBS4c9J4/DN7NuUMg4/AVz8o/Dn3Vuy+W1m1v4ZmZ1XT7T1hW+mVmdK3wzs94Q/d0dxLzQmLYN+/eU9KykzzSkOaatmXW2Ll8Pv+iYtnVfA66vP3FMWzOrhC5fHrnQmLYAko4HHiILg1jnmLZm1vlqkX+roEJj2kqaSDY29EsDjm8qpq1DHJrZqHCXzqCGimn7RbLlOp9t5aQiYl5ETI+I6WPGbNdKVmZm+XV5hV90TNtDgBMk/W9gJ6Am6QWyxflzx7Q1MxsVXT5KZ0QVvqRdI2L9wJi2EfHWhmO+CDwbEd9KcW6npjUi1pGtB/HBVk/ezKxQFe2bz6vomLaDiog+x7Q1s45X0dE3eQ1b4UfESUPs+sYwr/vigOfXAdflPjMzs3br9Ra+mVmviIrejM3LFb6ZWZ1b+GZmPcKjdMzMeoS7dMzMeoS7dMzMekSvD8s0M+sZbuGbmfWG6PNNWzOz3uAWvplZj3AfvplZj+jyFn6hMW0lTZb0J0l3pu3bDa85MB2/RtI3JamcSzIzG5moRe6tisqIaftARByQto83pF8MnApMTdvAPM3MRlevhzhsNqbtYCS9CtghIpZERABXAMc3f7pmZiXq68+/VVChMW2TKZKWS/qFpHpAlElkcWzrHNPWzDpPr7fwhzBUTNtHgT0j4o3AWcAP6v37zXBMWzMbDRGRe6uiQmPaRsRGYGN6fLukB8i6f9aRxbGtc0xbM+s8FW255zWiFr6kXdP/L4lpK+kVksamx68luzn7YEQ8CjwjaUYanXMKcG0B529mVpwu79IpOqbt24AvS3oRqAEfj4j6Dd/TyUb8TACuT5uZWceo6nDLvNTpfVHjxk/q7BM0s47Qt2ldy3N7nj75iNz1zY7f/Wnl5hJ5pq2ZWdLtLXxX+GZmda7wzcx6RHevneYK38ysrtu7dEY68crMrOtEX+TehjPEwpP/LOleSSslXSNpp4Z9c9PikvdJOroh/ZiUtkbSnIb0KZKWpvQfSRo/3Dm5wjczq6s1sQ3vcv5ykcibgH0jYn/gfmAugKRpwInAPuk1F0kam+Y1XQgcC0wDTkrHApwHXBARewEbgNnDnZArfDOzJGr5t2HzGmThyYi4MSL60tMlbF6BYCawMCI2RsRDwBrg4LStiYgHI2ITsBCYmSawHg5clV6/gBwLUrrCNzOra6KF37jIY9pOa7K0j7J5Auok4JGGffUFJodK3wV4quHLY4sLUtb5pq2ZWdJMhMOImAfMG0k5kj4P9AHfH8nrR8oVvplZXRuGZUr6MPAe4IjYvNTBOl66zHzjApODpT8B7CRpXGrl51qQ0l06ZmZJrS//NhKSjgE+BxwXEc837FoEnChpa0lTyBaevBVYBkxNI3LGk93YXZS+KG4GTkivn0WOBSkLjWmb9u2f9q1K+7dJ6Y5pa2Ydrcibtmnhyd8Ar5O0VtJs4FtkMURuaoz7HRGrgCuB1cANwBkR0Z9a758AFgP3AFemYwHOBs6StIasT//SYc9puMXTJL0NeBa4IiL2TWnLgM9ExC8kfRSYEhH/IGkccAdwckSskFS/sdAv6VbgTGApcB3wzYgYdsVML55mZnkUsXjaY4cdlru+2e3nP69co7XomLZHASsjYkV67ROpsndMWzPreEW28DtR0TFt9wZC0mJJd0j6XEp3TFsz63hRU+6tioqOaTsOOBT4UPr/vZKOaDZzx7Q1s9HQ7S38QmPakrXcfxkRj6d91wFvAr6HY9qaWYer9Vez5Z5XoTFtye4k7ydp23QD9+3Aase0NbMq6PYunUJj2kbEBklfIxs7GsB1EfGTdJxj2ppZR+vwiK8tc0xbM+sKRQzL/P2b3pm7vnnNHf9euWa+l1YwM0uq2lWTlyt8M7Ok22/ausI3M0siXOGbmfWEqo6vz8sVvplZUnML38ysN7hLx8ysR3iUjplZj/AoHTOzHuE+fDOzHtHtffiFhjiU9KEUtqu+1SQdkPY5xKGZdbSI/FsV5Vkt83LgmAFplwBzImI/4BrgswAR8f2IOCAiDgBOBh6KiDvTay4GTiULzjt1kDzNzEZVLZR7q6KiQxw2OglYCOAQh2ZWBRHKvVVR0SEOG/0N8MP0uKkQh2Zmo6G/ptxbFRUd4hAASYcAz0fE3YO9eDiOaWtmo6HbW/hFhzisO5HNrXvIwhnmDnEYEfOAeeD18M2sfaraN59X0SEO62kfIPXfAzjEoZlVQTSxVVGhIQ6TtwGPRMSDA7JyiEMz62jd3sJ3iEMz6wpFhDj8f688IXd989Y/XFW5bwfPtDUzS4LK1eFNcYVvZpbUurw/wRW+mVlScwvfzKw3uEvHOlrZP55bjxtfcgkQJQ9y23rsVqXmDzCm5LUA+2vlBlv9z01/KjX/qujykLau8M3M6vrdwjcz6w1u4ZuZ9Qj34ZuZ9YiKLoKZmyt8M7Ok24dljnR5ZDOzrtPfxJaHpL+TtErS3ZJ+KGkbSVMkLU3hXn8kaXw6duv0fE3aP7khn7kp/T5JR4/0+oqOabuVpAUp/R5Jcxtec0w62TWS5oz0hM3MylKTcm/DkTQJOBOYHhH7AmPJlo4/D7ggIvYCNgCz00tmAxtS+gXpOCRNS6/bhyw07EWSxo7k+gqNaUsW/WrrlH4g8DFJk9PJXQgcC0wDTkoXYWbWMUpYHnkcMEHSOGBb4FHgcOCqtH8Bm8O9zkzPSfuPSMvJzwQWRsTGiHgIWAMcPJLrKzqmbQDbpYubQBYJ65l0cmsi4sGI2ES2Vv5MzMw6SK2JrTEyX9pOa8wrItYB5wMPk1X0TwO3A09FRF86rDHc6yTgkfTavnT8Lo3pg7ymKUXHtL0KeI7s4h4Gzo+IJ2nyhB3i0MxGQ035t4iYFxHTG7Z5jXlJ2pmsnpwCvBrYjr/sLWmromPaHkx2P+PVZBf5aUmvbTbzxjdyzJjtRniKZmbNqaHcWw7vBB6KiD9GxItkwaLeAuyUekHgpeFe15Eaz2n/jsATjemDvKYpI6rwI+LeiDgqIg4ki137QNr1QeCGiHgxItYDvwamF3nCZmZl6Vf+LYeHgRmStk198UcAq4GbgRPSMbPYHO51UXpO2v+zyCJULQJOTKN4pgBTgVtHcn1Fx7R9mOyGBJK2A2YA9wLLgKlpONJ4sjvOi0ZStplZWZrpwx9ORCwl6+a+A7iLrL6dB5wNnCVpDVkf/aXpJZcCu6T0s4A5KZ9VwJVkXxY3AGdERN6RoS8xbIjDxpi2wGOkmLZAY0zbuRERkiaSxbedRraQ42UR8c8pn3cBXycbmjQ/Ir6a5wQd4nDLvFrm8Lxa5vC6YbXMIkIcXjbpv+X+YfzIuu9VbpbWsDNtI+KkIXZ9Y5BjnyW7iTtYPtcB1zV1dmZmbeSlFczMeoRXyzQz6xGu8M3MekTO0TeV5QrfzCxxC986mkoeHTJhq/JH6UwYW24Zb5z4mlLzB9hvzA6l5r9DlLuw7TmP3lxq/lXR7UMCXeGbmSUepWNm1iPcpWNm1iNGNH21Qlzhm5kl7tIxM+sR7tIxM+sR3T5Kp+iYtuMlXZbSV0g6rOE1B6b0NZK+qbLHE5qZNalG5N6qqOiYtqcCpPQjgX9JSygDXJz2T03bqEZ+MTMbqMjlkTtR0TFtpwE/S69bDzwFTJf0KmCHiFiSFvS/gs2Be83MOkJ/E1sVFR3TdgVwnKRxKTLLgWnfJLI4tnWOaWtmHaeZmLZVVHRM2/lklfltZMFObmEEX4aOaWtmo6Hb+/BHNEonIu4FjgKQtDfw7pTeB/xd/ThJtwD3AxvI4tjWOaatmXWcalbj+RUa0zYF690uPT4S6IuI1RHxKPCMpBlpdM4pbA7ca2bWEbr9pu2wLfzGmLaS1pJi2kpqjGl7WXq8K7BYUo2sBX9yQ1ank434mQBcnzYzs45R1a6avIqOafs74HVD5HMbsG8zJ2dm1k5VHX2Tl2fampklPd/CNzPrFd1d3bvCNzP7s6rejM3LFb6ZWRJd3sZ3hV9xtSj3B/TpF8qf6fzsmBdKzX9p7YFS8wc4evsDS83/1OVfLjX/c1791lLzr4o+V/hmZr2hu6t7V/hmZn/mUTpmZj3CN23NzHqEb9qamfUIt/DNzHpEf5e38PPEtN1D0s2SVktaJemTKf1lkm6S9Nv0/84pXSlm7RpJKyW9qSGvWen430qaVd5lmZk1rxaRe6uiPMsj9wGfjohpwAzgDEnTgDnATyNiKvDT9BzgWDbHrT2NLJYtkl5GttLmIcDBwBfqXxJmZp0gmtiqKE9M20cj4o70+D+Be8jCE84EFqTDFrA5Ru1M4IrILAF2SjFtjwZuiognI2IDWSxcBzI3s47hiFcNJE0G3ggsBXZLgU0A/gDslh5PAh5peFk9fu1Q6YOVcxrZXwdo7I44zKGZtUO3j9LJHfFK0kTg/wKfiohnGvdFRKF/5TimrZmNhqIjXkkaK2m5pB+n51MkLU33OH8kaXxK3zo9X5P2T27IY25Kv0/S0a1cX64KX9JWZJX99yPi6pT8WOqqIf2/PqWvA/ZoeHk9fu1Q6WZmHaGfWu4tp0+SdYPXnQdcEBF7kcX6np3SZwMbUvoF6TjS/dITgX3IusAvkjR2pNeXZ5SOgEuBeyLiaw27FgH1kTaz2ByjdhFwShqtMwN4OnX9LAaOkrRzull7VEozM+sIRbbwJe0OvBu4JD0XcDhwVTpk4L3P+j3Rq4Aj0vEzgYURsTEiHgLWkA16GZE8ffhvIYtNe5ekO1PaOcC5wJWSZgO/Bz6Q9l0HvCud2PPARwAi4klJXwGWpeO+HBFPjvTEzcyKFk0Mt2y815jMi4h5Dc+/DnwO2D493wV4KiL60vPG+5h/vscZEX2Snk7HTwKWNOQ55L3PPPLEtP0VoCF2HzHI8QGcMcixRMR8YH4zJ2hm1i7NjL5Jlfu8wfZJeg+wPiJul3RYMWfXOs+0NTNLClxa4S3AcZLeBWwD7AB8g2yY+rjUym+8j1m/x7lW0jhgR+AJCr73mXuUjplZtyvqpm1EzI2I3SNiMtlN159FxIeAm4ET0mED733W74mekI6PlH5iGsUzhWxC660jvT638M3Mkmb68EfobGChpH8ClpMNiCH9/11Ja4Anyb4kiIhVkq4EVpOtenBGRPSPtHC14QJbMm78pM4+QTPrCH2b1g11rzG3o/c4Nnd9s/iR61sur93cwjczS7p9pq0rfDOzpKpr5OTlCt/MLOn0Lu5WucI3M0uaWDKhklzhm5klVQ1skpcrfDOzpLur+3JCHL5e0m8kbZT0mQF5HZOW+Fwjac5g5ZmZjZZuD4BSRojDJ4EzgfMbM0lLel5IFgJxGnBSysfMrCP0fIXfbIjDiFgfEcuAFwdkdTCwJiIejIhNwMKUh5lZR+iPWu6tippaSydniMOh5A5xaGY2GqKJf1WU+6btwBCH2dr8mYgISYW9A45pa2ajodvH4ZcR4nAouZf5dExbMxsNPd+HP4IQh0NZBkxNQXzHk60Gt6j5UzYzK0dE5N6qqPAQh5JeCdxGtuB/TdKngGmpG+gTZHFsxwLzI2JVoVdjZtaCqrbc8/LyyGbWFYpYHnnf3Wbkrm/ufmyJl0c2M6uqqo6+ycsVvplZ4rV0zMx6hFv4ZmY9wi18M7Me4Ra+mVmPqOoaOXm5wjczS8IVvplZb+j2iVeu8M3Mkk6fiNoqV/hmZolb+GZmPaK/1t19+GXEtP2QpJWS7pJ0i6Q3NOTlmLZm1rG6PQBKGTFtHwLeHhH7AV8B5oFj2ppZ5+v25ZHLiGl7S0RsSOlLyAKdgGPamlmH6/YAKE314Y8gpu1s4Pr0eLCYtoc0U76ZWZmq2nLPq7SYtpLeQVbhH9rsSTmmrZmNBq+lw5Zj2kbEowNj2kraH7gEODYinkjJTcW0JfX9OwCKmbVLty+tUHhMW0l7AlcDJ0fE/Q3HO6atmXW0br9pW3hMW+AfgV2Ai1K3T19ETI+IPse0NbNO1u1dOo5pa2ZdoYiYttttOzl3ffPc879zTFszs6rq9hZ+nolXZmY9oeg+/E5bXcAtfDOzpFbgKJ2G1QWOJJt3tEzSoohYXVghTXIL38wsKbiF33GrC3R8C7/ZGzGSTkvj+EtR9fzbUUbV829HGb6G0c9/MC82Ud80ThBN5g04345bXaAbW/inDX9IT+ffjjKqnn87yvA1jH7+LYmIeWnIeX1r65fTSHRjhW9m1glyry7QLq7wzczK0XGrC3R8H/4IlP1nVdXzb0cZVc+/HWX4GkY//1J14uoCHT/T1szMiuEuHTOzHuEK38ysR7jCNzPrEa7wzcx6RDeO0jHrOZJ2BI4hm90J2XjvxRHxVIFlvJ5saYDGMhZFxD1VyN88SmdY7fgh9A/66KvyZyDpFOALwI1sntizO9miXV+KiCsKKONs4CSy9WDWNpRxIrAwIs7t5PwtU/kKX9LRwPG89Bf12oi4oYC8S/8hbFMZpb1H7ci/7DKq/hlIug84ZGBrXtLOwNKI2LuAMu4H9omIFwekjwdWRcTUTs7fMpWu8CV9HdgbuIKX/qKeAvw2Ij7ZYv6l/xC24Rep7Peo1PzbUUYXfAb3AwdFxNMD0ncEbivo5/Re4OiI+P2A9NcAN0bE6zo5f8tUvcK/f7DWSwq8fn8Bv6il/xC24Rep7Peo1PzbUUYXfAazyGJJ38jm1Rn3JOvS+UpEXN5K/qmMY4BvAb8dUMZewCda/Uul7PwtU/Wbti9IOigilg1IPwh4oYD8PwX8VNKgP4QF5N+OMsp+j8rOvx1lVPoziIgFkhYBR7O5y+jnwNyI2NBq/qmMGyTtTbbGe2O31LKI6O/0/C1T9Rb+m4CLge3Z/KfyHsDTwBkRcXsBZYyh5B/CMsso+z1q02dwIHBRyWVU9jNoKGc3Gs4/Ih4rIt8c5U6MiGfbUZa1ptIVfp2kV/LSH/Q/FJj3GICIqKU+3X2B30XEk0WVMUiZp0fERQXnWdp71I78yy5D0p7AMxHxlKTJwHTgniIXuyrr/CUdAHwb2JHsC0Vk9wieAk6PiDuKKGcL5T8cEXu2mMf+ZIulTQKuB86u/3Ui6daIOLj1M7Wqd+nUPTHwl0fSyyPi8VYylXQ88B2gJunjwDnAs8DrJP1tRPxbK/mnMs4aJPkcSdsARMTXWi0j5fMH4A+SJgJ7S3qhiDHa6UvwxYb83wEcJmlVgaOA9o+IlfUyishzQP5zgI8BGyWdD3wG+DXwJUmXFvUZkFXCewD9wHMUdy2XAx+LiKWNiZJmAJcBb2i1gCF+TiH7cpnYav5kf8F9EVgC/HfgV5KOi4gHgK0KyN+o+ExbSe+QtBZ4VNKNqWVWd2MBRXyB7JflzcB3gVMi4gjgLWlfEb5EFvZsItmf/NuTLaVaf9wSSRc1PD4UWA38C3CXpHe1mj/Zmt87pfw/C3wVmAB8WtL/KiB/gOWSfivpK5KmFZRno5OBaWSf6wXAWyNiNlkXz0dbzVzS2yXdBpwLzCeL5HSppJ9L2mPLr85lu4GVPUBELAG2KyB/gP8J7Mzmn8v6NpFi6pHtI+KGiHgqIs4nu3dyQ/rSqn43RKeIJoL2dtpGVtnskx6fQHaHf0Z6vryA/Jc3PL57wL47CrqGPYH/A5wHbJvSHizwPbqj4fHNwJvS49eSDdlrNf+7Gx7fBkxIj8cBKwu6huVkXWlfBdYAK4A5wOSC8l+Z/h8LrAfGDPW5t3D+r0iPpwDXpMdHko0CajX/bwI/Af6GrHHy5vT4J8C3CnqPbgEOHGLfIwXkvwLYcUDa/ul3+okirsFbVL5LZ3ykPtaIuErSPcDVaSJNIa0CSWMiokZDS0/SWGB8EflHxMPA+yXNBG6SdEER+Q5hh0j9uRHxYP3+RIuekbRvRNwNPA5sA/yJrMIv6i/ISPl/Hvi8pIPJJkX9KvUfv7nF/O+Q9AOy1vBPgQWSbgAOJ/uLqFVjI+KP6fHDwGsAIuKmNEa/JRFxpqRj+cuZwhdGxHWt5p98BBjqvtX0AvI/D/ivZF06AETESklHAP9QQP5GxW/apj+T3xMN/feSdgd+DPyXiGipS0TSQcBdEfHCgPTJwKER8b1W8h+kvIlkXUWHRMTbCsrzebJWsYDJwJ4RsSFV9isjYt8W89+frLtrRUp6C/BLYD/gaxHxg1byT2Usj4g3DpIu4G0R8YsW8x8HvJ+skXAVWRfbSWSV84UR8VyL+c9Pef8MOI7shu1ZkrYl+wvs9a3kb5ZX1Sv8dwJ/jIgVA9J3Ihvu9tXRObPOkSYPNXo0IjZJejlZZXl1AWWMBY4im006jmykSGELd0n6YBFfHKNF0lbAqWT3CVaQhbrrlzQB2DUGTPgaQf47AnPJWvi7kX25rAeuBc4t4nNoKON4YNeiyyg7f8tUusIvW2pxfw54H9kIi03AA8C3o4DZi+0qw7ZsC5/BxRGxYDTPLQ9Ji8n+elhQ/2s3DQH9MHB4RBxVYhmzgCNaLaPs/C1T6Qq/7MpS0rXANcC/Ax8g6+NdCPw92Z/l53R6GW14jyr/pdjGz+CvyYZlFn3+98UQyz9saV8nldGOa7DqV/hl/6KuiIg3NDxfFhEHpf7v1UX0vZZdRhveo274Uqz6Z3BjyntBpNm1ymbdfhg4MiLe2Ur+7SijHddgVH5Y5ooBz5el/8cA9xaQ/y1kN2chu9m2uGHffQVdQ6lltOE9KjX/bvic23D+O5ONcrkX2EA2muaelPaygj6DUstoxzV4q/6wzOckHRoRv5J0HGnYWGTLIKiA/D8OXCJpKrCKNDRT0iuACwvIvx1llP0elZ1/O8qo9GcQ2airy4CbgCXRsK6NslUoW57xXHYZ7bgGo/It/P2BW8laBL8C9k7prwDOLLnsj7Th+louo+z3qB2fQdU/5zZ8BmcC9wH/CvwOmNmwr6gJgqWW0Y5r8BbVrvC3eGElV8jAw224hlLLaMN7VIkvxap/BsBdwMT0eDLZjOdPpuctzzhvRxntuAZvUe2btluiYlbwWznULrJOMfYzAAAA9klEQVRW2tat5N+uMrZQdsvv0WjmX1QZVf8MlC1Ut0/D84lkE8hWkw3LPKDF0yy9jHZcg1V8tcxhflF3K6CI3ciCSgwMIiGyG31FKLWMst+jNnwGlf+c23D+j0k6ICLuBIiIZyW9h2yhtv0KyL8dZbTjGnpepSt8yq+Qf0z2Z+adA3dI+nkB+bejjLLfo8p/KVL9z+AUoK8xISL6gFMkfaeA/NtRRjuuoedVvcIv9Rc1siVyh9r3wVbzb1MZZVdmlf9SrPpnEBFrt7Dv163m344y2nENVvGJV2Zmll+lA6CYmVl+rvDNzHqEK3wzsx7hCt/MrEf8f874PetpcC+7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "ind = list(range(1920,2030,10))\n",
    "df_heat = pd.DataFrame(cm, index=ind, columns=ind)\n",
    "len(ind)\n",
    "# lab = pd.unique(df_heat[0])\n",
    "sns.heatmap(df_heat)\n",
    "df_heat\n",
    "# df_plot.transpose().corr()\n",
    "# cm.shape"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}