Python_for_Data_Science/simple_ml.py at master · gumption/Python_for_Data_Science

Abstraction: Python utility library implementing decision tree ML from scratch

Key points:

• Author Joe McCarthy (v1.0.3, 2014); ~383 lines implementing simple ML tasks in pure Python
• Implements create_decision_tree via simplified ID3 algorithm — recursive splitting by highest information gain
• Provides entropy, information_gain, majority_value, classify, and classification_accuracy functions
• Supports loading/saving CSV instances and attribute name/value metadata files
• Includes compute_learning_curve for evaluating model accuracy across incrementally larger training sets

Connections: Github · Decision Trees · Information Gain · Machine Learning

Source: https://github.com/gumption/Python_for_Data_Science/blob/master/simple_ml.py