Mine-Detection-Neural-Network-Unsupervised-MLP-PCA

Using a neural network, a Multi-Level Perceptron, to train and test on the Kaggle Sonar, Mines vs. Rocks Dataset.

• The Mines vs Rocks dataset on submarine is designed where the rightmost column is a string, either "R"(Rock) or a "M"(Mine)
• The dataset is also non-labled for the inputs requiring for Unsupervised Learning, the Primary Component Analysis. This is ran to predict which input components cause a difference in the dataset when creating a model for it.

Requirements

• Python 3

Packages

• numpy version 1.16.5 or +
• pandas version 1.0.0 or +
• scipy version 1.3.1 or +
• matplotlib version 1.3 or +
• scikit-learn version 0.21.3 or +
• seaborn version 0.9.0 or +

• I used the Anaconda Environment to install these packages with additional, with Jupyter Notebook and Spyder IDE in addtion
• Other method to just download packages is Miniconda