Repository contains a range of from-scratch python functions for Machine Learning. Included functions range from preprocessing techniques such as bootstrapping to full object-oriented classification models such as K-Nearest Neighbours (KNN). They are written such that they are easily adaptable to a range of input forms and only use NumPy for array manipulation and SKlearn for induvidual decision trees.
Functions include:
- Rebalence and Stratified Split
- Classifier Performance Metrics
- Logistic Regression
- K-Nearest Neighbours Classifier and KNN Variants e.g. Cost-Sensitive, Grouped, Auto-Grouped
- Bootstrapping Function
- Adaptive Boosting Classifier (AdaBoost)
- Random Forest Classifier
- OvO Classification Algorithm