The-Fugazi/Nigeria-anomaly-detector

Detects anomalies in Nigeria's USD/NGN exchange rate using Isolation Forest, Z-Score, and Rolling Deviation. Built with Python, Scikit learn, and Streamlit. # 🇳🇬 Nigeria Exchange Rate Anomaly Detector A data engineering and machine learning dashboard that detects anomalies in the USD/NGN exchange rate using three detection methods: **Isolation Forest**, **Z-Score**, and **Rolling Deviation**. ## Live Demo View on Streamlit Cloud ## What it does - Pulls historical CBN USD/NGN exchange rate data - Runs three anomaly detection algorithms in parallel - Flags and labels major economic shock events (2016 devaluation, 2020 COVID crash, 2023 naira float) - Visualizes anomalies interactively with hover context and % change - Allows date filtering and CSV export of flagged anomalies ## Key Events Detected | Date | Event | Rate Change | |------|-------|-------------| | Jul 2016 | CBN devaluation | ₦197 → ₦305 (+55%) | | Jul 2020 | COVID-19 crash | ₦360 → ₦461 (+28%) | | Jun 2023 | Tinubu naira float | ₦463 → ₦770 (+66%) | | Dec 2023 | Further depreciation | ₦1,490 | ## Tech Stack - **Python** — data pipeline and ML - **Streamlit** — dashboard and deployment - **Scikit-learn** — Isolation Forest - **Plotly** — interactive charts - **Pandas/NumPy** — data wrangling ## Run locally ```bash git clone github.com cd nigeria-anomaly-detector python -m venv venv venv\Scripts\activate pip install -r requirements.txt streamlit run app.py ``` ## Author **Allison Oluwatobi Samuel** — Data Analyst & ML Engineer GitHub · LinkedIn