Santandave961/pos-analyzer-nigeria

# POS Analyzer Nigeria - POS Transaction Dashboard + ML Anomaly Detection A data science web application that analyzes Nigerian Point of Sale (POS) transaction patterns across states and merchant categories, with Isolation Forest machine learning anomaly detection. Built with Python and Streamlit, deployed on Streamlit Community Cloud. --- ## Live Demo [Click here to view the app]pos-analyzer-nigeria-7esqtz… --- ## Overview This project builds an interactive analytics dashboard that gives visibility into Nigerian POS transaction behaviour across 10 states, 10 merchant categories, and 8 banks — while using an unsupervised machine learning model to automatically flag suspicious transactions in real time. --- ## Model Performance | Metric | Value | |--------|-------| | Algorithm | Isolation Forest | | Estimators | 200 trees | | Contamination | 5% | | Scaler | StandardScaler | | Dataset | 500 synthetic Nigerian POS transactions | | Anomalies Injected | 25 (5%) with amounts 8-20x normal | --- ## Features **Dashboard:** - Total transaction volume, count, average, and anomaly rate KPIs - Filter by State, Category, and Bank - Transaction volume and count by Nigerian state - Spending breakdown by merchant category - Volume comparison by bank - Monthly transaction trend - Hour of day activity chart - Weekday vs Weekend analysis **ML Anomaly Detection:** - Isolation Forest trained on 500 transactions - Anomalous transactions scatter plot over time - Anomalies by state and category - Flagged transactions table - Single transaction checker — input any transaction and get instant Normal or Anomaly verdict --- ## States Covered Lagos, Abuja, Kano, Rivers, Oyo, Kaduna, Enugu, Delta, Anambra, Ogun --- ## Merchant Categories Food and Beverage, Transport, Retail, Healthcare, Entertainment, Utilities, Education, Fuel, Fashion, Electronics --- ## Banks Covered GTBank, Access Bank, First Bank, Zenith Bank, UBA, Kuda, Opay, Mon …