Decision Tree Algorithm Project

Overview

This repository contains the implementation of a Decision Tree algorithm for classification tasks, applied to two datasets:

1. Fraud Detection: Classifies individuals as "Risky" or "Good" based on taxable income and other attributes.
2. Sales Analysis: Identifies attributes contributing to high sales for a cloth manufacturing company.

---

Datasets

1. Fraud Detection Dataset:

   • Attributes: Undergrad, Marital.Status, Taxable.Income, Work Experience, Urban.
   • Target: "Risky" (Taxable.Income ≤ 30,000) or "Good".

2. Sales Dataset:

   • Attributes: Sales, CompPrice, Income, Advertising, Population, Price, ShelveLoc, Age, Education, Urban, US.
   • Target: Categorical variable derived from Sales.

---

Project Structure

📂 Decision-Tree-Algorithm
├── 📁 data                # Dataset files
│   ├── fraud_data.csv
│   ├── sales_data.csv
├── 📁 notebooks           # Jupyter notebooks for data exploration and model building
│   ├── fraud_detection.ipynb
│   ├── sales_analysis.ipynb
├── 📁 scripts             # Python scripts for modular code
│   ├── preprocess.py      # Data preprocessing functions
│   ├── train_model.py     # Model training and evaluation
│   ├── visualize_tree.py  # Decision tree visualization
├── 📁 results             # Outputs such as visualized trees and metrics
│   ├── fraud_tree.png
│   ├── sales_tree.png
├── README.md              # Project overview
├── requirements.txt       # Python dependencies
├── LICENSE                # License details
└── .gitignore             # Files to ignore in Git

---

Key Features

1. Fraud Detection Model:

   • Accuracy on Training Set: 1.0
   • Cross-Validation Score: 0.998
   • Visualized decision tree for interpretability.

2. Sales Analysis Model:

   • Identifies critical factors driving high sales.
   • Feature Importance:
     • Advertising: 16.01%
     • Population: 16.72%
     • ShelveLoc: 12.25%

---

Setup

1. Clone this repository:

   git clone https://github.com/R-Mahesh45/Decision-Tree-Algorithm.git
   cd Decision-Tree-Algorithm

2. Install dependencies:

   pip install -r requirements.txt

3. Run the scripts:
   • For fraud detection:

     python scripts/train_model.py --dataset data/fraud_data.csv

   • For sales analysis:

     python scripts/train_model.py --dataset data/sales_data.csv

---

Results

• Fraud Detection Tree:

• (https://github.com/user-attachments/assets/86612b54-c1dc-463f-8a1a-d993e31db6df)

• Sales Analysis Tree:

• (https://github.com/user-attachments/assets/deaa34a1-cb4a-4cfd-a54e-d041a26abec7)

---

Visualization

• Visualize decision trees:

  from sklearn.tree import plot_tree
  import matplotlib.pyplot as plt
  
  plt.figure(figsize=(15, 10))
  plot_tree(model, filled=True, feature_names=X.columns, class_names=["Risky", "Good"])
  plt.show()

---