Comparative Study: Decision Trees vs. Linear Models

This repository explores the performance trade-offs between Tree-based models and Linear models across two distinct domains: Healthcare (Classification) and Automotive Engineering (Regression).

Comparison Summary

Task	Dataset	Model A (Linear)	Model B (Tree)
Classification	Wisconsin Breast Cancer	Logistic Regression	Decision Tree Classifier
Regression	Auto-MPG	Linear Regression	Regression Tree

---

Project 1: Breast Cancer Diagnostic (Classification)

Goal: Classify tumors as Benign (0) or Malignant (1).

• Core Strategy: Comparing the linear decision boundary of Logistic Regression against the non-linear "splitting" logic of a Decision Tree.
• Result: Even using only 2 features (radius_mean and concave points_mean), the Decision Tree identifies complex patterns that a straight line might miss.

Project 2: Auto-MPG Efficiency (Regression)

Goal: Predict a vehicle's miles-per-gallon (MPG) consumption.

• Core Strategy: Testing if automotive performance follows a linear trend or a hierarchical structure.
• Key Discovery: Using 6 technical features, the Regression Tree captures non-linear interactions (e.g., how horsepower and weight combined affect fuel) more flexibly than standard Linear Regression.

---

Environment Setup (Conda)

1. Create the Environment

git clone https://github.com/Joe-Naz01/ml_tree.git
cd ml_tree

conda create -n ml_tree
conda activate ml_tree

pip install requirements.txt