UCI Thyroid Classification - Python, Keras, scikit-learn, ANN

This project is created for classification problem on UCI-Thyroid-Disease dataset. It uses ANN to make predictions. Prediction classes are:

• 1-Hyperthyroid
• 2-Subnormal
• 3-Normal

Data Sets

• Thyroid disease data set in UCI repository.

Frameworks/Libraries

• Keras
• scikit-learn

Getting Started

These instructions will get you a brief idea on setting up the environment and running on your local machine for development and testing purposes.

Prerequisities

• python3.5 or newer
• Keras
• scikit-learn
• numpy
• pandas

Setup and running tests

1. Run python -V to check the installation

2. Install all the required libraries.

3. Execute the following commands from terminal to run the tests:

   python main.py

Note: Model accuracy - on validation: 98.57% (0.41%) and on test data: 98.279%. As per the dataset information any model with accuracy >92% is considered as a good one. Further improvement is definitely possible.