SVR Decision Tree Multiclass Classifier This repository contains the implementation of the SVR Decision Tree Multiclass Classifier, a novel approach designed to address the challenges posed by imbalanced datasets in multiclass classification problems. The classifier leverages the Surface-to-Volume Ratio (SVR) regularization technique to transform the decision boundaries of decision trees, improving their performance on imbalanced datasets.

Introduction Many classification algorithms struggle when faced with real-time data characterized by imbalanced class distributions in the target data. Traditional decision trees, while interpretable and effective, can be unstable and biased towards the majority class. To address these issues, we propose the SVR Decision Tree Multiclass Classifier, a novel approach that enhances decision tree performance on imbalanced datasets.

Our classifier employs the Surface-to-Volume Ratio (SVR) regularization technique to adjust the decision boundaries of individual decision trees. This regularization aims to transform irregularly shaped decision boundaries, leading to improved classification accuracy, particularly on imbalanced multiclass datasets.

Features Utilizes the Surface-to-Volume Ratio (SVR) regularization technique. Enhances decision tree performance on imbalanced multiclass datasets. Improved classification accuracy on irregularly shaped decision boundaries. Easy-to-use interface for integration into existing machine learning pipelines. Comparative evaluation against traditional and recent imbalanced classification algorithms.

Paper Link: https://www.doi.org/10.56726/IRJMETS32590