📘 ARFF Tree Explorer

A Python-Based Machine Learning Visualization Tool for ARFF Datasets

---

🚀 Overview

ARFF Tree Explorer is a Python-based desktop application designed to load ARFF datasets, train multiple classification models, and visualize decision trees in an intuitive, beginner-friendly GUI.

The tool makes learning ML more accessible by combining:

• Python GUI (Tkinter)
• ARFF file processing
• Decision tree visualization
• Classifier performance comparison
• User-friendly workflow for education & research

This system acts as a lightweight, open-source alternative to Weka.

---

🎯 Key Features

🔹 Classifier Comparison Module

• Load ARFF files
• Automatic preprocessing
• Evaluate models using 5-fold cross-validation
• Compare accuracy of:
  • J48 (Entropy-based Decision Tree)
  • REPTree
  • Random Forest
  • Decision Stump
  • Logistic Regression (LMT Approximation)

---

🔹 Decision Tree Visualization Module

• Select model from dropdown
• Train automatically
• Visualize decision trees using Matplotlib
• Explore splits, nodes, and decision paths

---

🔹 Core System

• Built fully in Python
• Handles categorical encoding
• Model training + evaluation in real-time
• Simple, minimal, student-friendly interface

---

🧩 System Architecture

User GUI (Tkinter)
        │
Data Processing (Pandas + ARFF Loader)
        │
Model Training (Scikit-learn)
        │
Classifier Comparison / Tree Visualization
        │
Matplotlib Output (Decision Trees)

---

🛠️ Tools & Technologies

Component	Technology
Language	Python 3.x
GUI	Tkinter
ML Models	Scikit-learn
Data Loader	scipy.io.arff
Visualization	Matplotlib
IDE	VS Code

---

📁 Project Structure

ARFF-Tree-Explorer/
├─ main.py
├─ assets/
│   ├─ screenshots/
│   └─ icons/
├─ sample_datasets/
├─ requirements.txt
└─ README.md

---

⚙️ Installation & Setup

1️⃣ Install Dependencies

pip install pandas scikit-learn matplotlib scipy

2️⃣ Run the Application

python main.py

---

🔍 How It Works

User Workflow

1. Upload ARFF file
2. System loads → preprocesses data
3. Compare classifiers
4. Visualize decision trees

Behind the Scenes

• ARFF parsed using scipy.io.arff
• Categorical features encoded with LabelEncoder
• Models trained via Scikit-learn
• Tree visualized using plot_tree()

---

🧠 Algorithms Used

📌 Classifier Comparison Algorithm

• Evaluates multiple ML models
• Uses 5-fold cross-validation
• Outputs accuracy scores

📌 Decision Tree Visualization Algorithm

• Trains selected classifier
• Generates complete tree plot
• Supports J48, REPTree, RandomForest, etc.

---

🗂️ Flowchart

---

📸 Demo Preview

🏠 Home Page

📊 Classifier Comparison Module

🌳 Decision Tree Visualization Module

🌲 J48 Tree Visualization

---

📈 Performance Evaluation

• Fast dataset loading
• Accurate ML model comparison
• Clear visual decision trees
• Smooth GUI interaction
• Tested on multiple UCI ARFF datasets

---

⚠️ Limitations

• Only ARFF supported (no CSV/XLSX)
• GUI may freeze with large datasets
• No threading / background tasks
• No exporting of tree images
• No hyperparameter tuning

---

🔮 Future Improvements

• Support CSV, XLSX, JSON
• Multi-threading for large data
• Save/Export trees
• Add hyperparameter tuning panel
• Zoomable, interactive tree viewer
• Dark mode for GUI

---

📄 Full Project Report (PDF)

(Complete documentation, flowcharts, case studies, screenshots)

👉 Download:
Download / View Project Report (PDF)

---

👥 Authors

• Md. Zehadul Islam
• Md. Abdullah Al Moin