Decision Tree (CART-like) + Tkinter Visualizer

Overview

This project contains a simple CART-style Decision Tree classifier implemented from scratch in Python, plus a Tkinter GUI viewer that renders the tree as a graph with:

• Black background
• White square nodes and edges
• White text
• Mouse-wheel zoom (Ctrl + wheel by default)

The goal is to keep the implementation clear and educational, while still being reasonably efficient (especially for numeric splits).

---

Features

Decision Tree

• Supports binary splits:
  • Numeric features: feature <= threshold
  • Categorical features: feature == value vs != value (one-vs-rest)
• Works with binary and multi-class targets
• Split criterion: Gini impurity
• Includes basic stopping rules:
  • max_depth
  • min_samples_split
  • pure nodes (gini == 0)

GUI Tree Viewer (Tkinter)

• Draws nodes as rectangles
• Draws edges as straight white lines
• Zoom in/out:
  • Ctrl + MouseWheel (Windows/macOS)
  • Ctrl + Button-4 / Button-5 (Linux)

---