👥 MBTI Text Classifier

A machine learning project that classifies text into one of the 16 MBTI personality types based on user posts. This project uses classic ML techniques with TF-IDF feature extraction to analyze text and predict personality types.

📋 Overview

This project implements a complete ML pipeline for MBTI personality classification:

• 📥 Downloads and preprocesses the MBTI dataset from Kaggle
• 🧹 Performs text cleaning and TF-IDF feature extraction
• ⚖️ Handles class imbalance using Random Oversampling
• 🤖 Trains and evaluates three classic ML models
• 📊 Provides detailed error analysis and visualizations

🎯 MBTI Types

The Myers-Briggs Type Indicator consists of 16 personality types based on four dimensions:

Dimension	Description
E/I	Extraversion vs Introversion
S/N	Sensing vs Intuition
T/F	Thinking vs Feeling
J/P	Judging vs Perceiving

📁 Project Structure

MBTI-Text-Classifier/
├── notebook.ipynb           # Main Jupyter notebook
├── requirements.txt         # Python dependencies
├── .pre-commit-config.yaml  # Pre-commit hooks
├── .gitignore
├── LICENSE
├── README.md
├── docs/
│   ├── REPORT_FA.md        # Project report in Persian
│   └── REPORT_EN.md        # Project report in English
├── data/
│   ├── raw_data.csv         # Original dataset
│   └── processed_data.csv   # Cleaned dataset
├── models/
│   ├── svm_model.pkl        # Trained best model
│   └── tfidf_vectorizer.pkl # TF-IDF vectorizer
└── images/                  # Generated visualizations
    ├── 01_mbti_distribution.png
    ├── 02_original_class_distribution.png
    ├── 03_resampled_class_distribution.png
    ├── 04_model_performance_comparison.png
    ├── 05_confusion_matrices.png
    ├── 06_per_type_accuracy.png
    └── 07_dimension_accuracy.png

🚀 Getting Started

Prerequisites

• Python 3.11+
• pip

Installation

1. Clone the repository:

   git clone https://github.com/Mr-MRF-Dev/MBTI-Text-Classifier.git
   cd MBTI-Text-Classifier

2. Create and activate virtual environment:

   python -m venv venv
   
   # Windows
   .\venv\Scripts\activate
   
   # Linux/Mac
   source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. (Optional) Install pre-commit hooks:

   pre-commit install

5. Run the notebook:

   jupyter notebook notebook.ipynb

   or using JupyterLab:

    jupyter lab notebook.ipynb

🤖 Models

Three classic ML models are trained and compared:

Model	Accuracy	F1-Score
SVM	~90%	~0.90
Logistic Regression	~88%	~0.87
Naive Bayes	~78%	~0.78

📊 Results

The SVM model achieves the best performance:

• ✅ Accuracy: ~90%
• ✅ F1-Score: ~0.90
• ✅ Best dimension: S/N (~98%)
• ⚠️ Hardest dimension: J/P (~94%)

📈 Visualizations

The notebook generates various visualizations saved in the images/ folder:

• MBTI type distribution
• Class distribution (before/after oversampling)
• Model performance comparison
• Confusion matrices for all models
• Per-type classification accuracy
• Dimension-level accuracy (E/I, S/N, T/F, J/P)

🛠️ Tech Stack

Category	Technologies
Language	Python
Data	Pandas, NumPy
ML	Scikit-learn, imbalanced-learn
NLP	NLTK, TF-IDF
Visualization	Matplotlib, Seaborn
Dataset	KaggleHub

🔧 Pre-commit Hooks

This project uses pre-commit for code quality:

• black - Code formatting
• isort - Import sorting
• flake8 - Linting
• nbqa - Notebook code quality

🙏 Acknowledgments

• Dataset: MBTI Dataset on Kaggle

🤝 Contributing

We welcome any contributions you may have. If you're interested in helping out, please fork the repository and create an Issue or Pull Request. We'll be happy to review your contributions.

📝 License

This project is licensed under the MIT License. For more information, please refer to the LICENSE file.