Ensembles of Oblique Decision Trees

Author: Torsha Majumder

Background

This repository contains several decision tree algorithms compatible with Scikit-Learn's Bagging Classifier. For the complete experimental setup and results, please check my thesis. If you find this code useful, please cite my work.

Citation

@mastersthesis{UTDthesis2020EODT,
  author       = {Majumder, T.},
  title        = {Ensembles of oblique decision trees},
  school       = {University of Texas, Dallas},
  year         = {2020},
  type         = {Master's Thesis},
  note         = {UTD Theses and Dissertations}
}

Experiment

Decision Trees considered for this experiment:

* Standard Decision Tree with Bagging
* Oblique Classifier 1 with Bagging
* Weighted Oblique Decision Tree with Bagging
* Randomized CART with Bagging
* HouseHolder CART with Bagging
* Continuous Optimization of Oblique Splits with Bagging
* Deep Neural Decision Trees with Bagging
* Non-Linear Decision Trees with Bagging
* Random Forest Classifier

In this experiment we have to skip OC1, DNDT and, NDT classifiers due to its computational cost.

This experiment has been conducted on 12 Benchmark Data sets.

* Iris                 * Vehicle
* Wine                 * Fourclass
* Glass                * Segment
* Heart                * Satimage
* Breast-cancer        * Pendigits
* Diabetes             * Letter