Crime Classification Project

Project Overview

The Crime Classification Project implements a machine learning system to classify crimes into six categories:
Murder, Rape, Assault, Body Found, Kidnap, Robbery.

The project aims to:

• Analyze crime data efficiently.
• Build predictive models to classify crimes accurately.
• Provide interactive visualization for insights using Streamlit.
• Deploy a user-friendly Flask web app for real-time predictions.

This project uses XGBoost and AdaBoost classifiers to achieve high accuracy and interpretability.

---

Key Features

• Data cleaning and preprocessing (cleaning.py)
• Model training, selection, and evaluation (training.py)
• Feature importance analysis
• Learning curve visualization
• Streamlit dashboard (crime_dashboard.py)
• Flask web app (app.py + index.html) for live predictions
• Model persistence (model.pkl, encoder.pkl)

---

Folder Structure

crime-classification/
│
├── cleaning.py
├── training.py
├── app.py
├── index.html
├── crime_dashboard.py
├── model.pkl
├── encoder.pkl
├── df.pkl
├── requirements.txt
├── README.md
└── assets/
    ├── dashboard.png
    ├── flask_app.png
    ├── confusion_matrix.png
    ├── feature_importance.png
    └── banner_screenshot.png

Installation

1. Clone the repository:

git clone https://github.com/yourusername/crime-classification.git
cd crime-classification

2. Make the virtual environment

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

Install the requirements.

pip install -r requirements.txt

Data Description

The dataset contains numeric and categorical features relevant to crime analysis. The target column is crime with six classes:

Crime Type	Label
Murder	0
Rape	1
Assault	2
Body Found	3
Kidnap	4
Robbery	5

Note: The project currently uses a sample dataset for demonstration, but all scripts are compatible with real crime datasets.

Usage

python cleaning.py
python training.py
streamlit run crime_dashboard.py
python app.py

Flask App Screenshot:

Dashboard Screenshots

Model Performance

Model	Accuracy	Precision	Recall	F1-score
XGBoost	0.93	0.93	0.93	0.93
AdaBoost	0.73	0.73	0.73	0.73

Future Work

• Integrate with a real-time crime dataset.
• Improve prediction accuracy with ensemble methods.
• Add geospatial analysis for crime hotspots.
• Deploy the Flask app on cloud platforms (Heroku, AWS).
• Add user authentication for secure usage.

Contributing

• Fork the repository
• Make your changes
• Submit a pull request

About the developer

Sarder Junaid Ahmed

Data Scientist & Machine Learning Engineer

Transforming complex data into strategic decisions through rigorous statistical modeling and production-ready machine learning systems.

Specializations: Statistical ML · Causal Inference · Trustworthy AI · Fairness-Aware ML · RAG Systems

Selected Research:

• 📄 Ahmed, S.J. et al. (2026). Machine Learning for Crime Classification: A Fairness-Aware Approach to Class Imbalance. Journal of Machine Learning and Applications, 2(1), 9–17. DOI: 10.61577/jmla.2026.100002
• 📄 Ahmed, S.J. et al. (2026). Machine Learning for Crime Classification: A Fairness-Aware Approach to Class Imbalance. IEEE SPICSCON 2026, BAUET, Bangladesh (Aug 13–14, 2026). Accepted for Presentation — IEEE Xplore.
• 📄 Ahmed, S.J. et al. (2026). CF-EGAT: A Causal Fairness-Aware Equity Graph Attention Network for Country-Level Environmental Livability Classification. SPECTRA 2026. 🏆 1st Best Paper Award
• 📄 Ahmed, S.J. (2025). Multi-Dimensional Statistical Similarity for Governance Classification: Beyond Arbitrary Thresholds. APMEE 2025. 🏆 Best Research Paper Award
• 📄 Ahmed, S.J. (2026). DeepEnMap: Ordinal-Aware Multi-Modal Deep Learning for Energy Poverty Risk Mapping. IEMIS 2026, University of British Columbia, Vancouver, Canada (Aug 10–12, 2026). Accepted for Presentation — Springer LNNS Series (Scopus, EI-Compendex, DBLP, ISI Proceedings).
• 📄 Ahmed, S.J. (2026). Density-Decoupled, Mask-Ablated Segmentation-Guided Diffusion for Controllable Mammography Synthesis: A Preliminary Study. IEMIS 2026, University of British Columbia, Vancouver, Canada (Aug 10–12, 2026). Accepted for Presentation — Springer LNNS Series (Scopus, EI-Compendex, DBLP, ISI Proceedings).
• 📄 Ahmed, S.J., Islam Nahian, M.T., & Kwoshik, M.H.R. (2026). Environmental Livability Assessment via Adaptive Bootstrap-Retrained SHAP and Statistically-Constrained Pareto Counterfactuals: A Cross-National Analysis. IEEE SPICSCON 2026, BAUET, Bangladesh (Aug 13–14, 2026). Accepted for Presentation — IEEE Xplore.
• 📄 Ahmed, S.J. (2026). DemocracyGuard: Testing a Divergence-Index Reconciliation of Subjective and Objective Democracy Indicators for Forecasting Adverse Regime Transitions. Under Review, Transactions on Machine Learning Research (TMLR) — Q1, Top-Tier Journal.
• 📄 Ahmed, S.J. (2026). FAI: Feature-Wise Adaptive Imputation via Downstream-Aware Method Selection. Under Review, ICISET 2026 (IEEE Xplore).

Other Deployed Projects:

• 🔬 ReproHub — Automated research reproducibility platform with composite scoring across 11 statistical tests
• 📊 StatsPro — AI-powered statistical analysis platform with automated CSV-to-report workflows

Honors: 🏆 1st Best Paper — SPECTRA 2026  ·  🏆 Best Research Paper — APMEE 2025  ·  🎖️ Esteemed Alumni Award — YLRL RUET 2024  ·  ⭐ Perfect GPA 5.00/5.00 — SSC & HSC  ·  🎓 National Merit Scholarship — 2009 & 2013

License

MIT License

Acknowledgements

• Scikit-learn
• XGBoost
• Streamlit
• Flask
• Plotly