mazimum86/nigeria-state-predictor

# 🌍 Nigerian State Predictor Using Latitude & Longitude This Streamlit web app uses a trained Artificial Neural Network (ANN) to predict the Nigerian state for any given **latitude and longitude**. It also returns the **top 3 most probable states** and their associated prediction confidence. > 🎯 Built for data-driven geospatial classification using major telecoms and collocation partner datasets across Nigeria. --- ## 📌 Demo 🔗 Live App: Streamlit Cloud Deployment --- ## ✨ Features - 📍 Predicts one of **37 Nigerian regions** (36 states + FCT renamed to Abuja) - 🔢 Uses a trained ANN (`main_model.keras`) and scaled features via `scaler.pkl` - 📊 Shows top 3 predictions with probabilities - 🧠 Trained on telecom infrastructure geocoordinates - 🌐 Powered by TensorFlow, Streamlit, and Scikit-learn --- ## 🧠 How It Works ### Input: - Latitude (e.g., `9.0578`) - Longitude (e.g., `7.4951`) ### Output: - Most likely state: `Abuja` - Probability: `92.4%` - Other candidates: `Nasarawa (6.3%)`, `Kogi (1.3%)` --- ## 🛠️ Setup Locally ```bash # Clone the repo git clone github.com cd nigeria-state-predictor # Create and activate a virtual environment conda create -n streamlit-env python=3.10 conda activate streamlit-env # Install required libraries pip install -r requirements.txt # Run the app streamlit run main.py ``` ## 📁 Project Structure ``` 📦 nigeria-state-predictor/ ┣ 📜main.py ┣ 📜requirements.txt ┣ 📜runtime.txt ┣ 📜scaler.pkl ┣ 📜main_model.keras ┗ 🖼️screenshot.png ├── .streamlit/ │ └── config.toml └── README.md ``` ## 📸 Screenshots ## 📤 Deployment App is deployed on Streamlit Cloud. Free to use, open source, and no signup required for visitors! ## 🙋‍♂️ Creator Info 🔧 Created by: Chukwuka Chijioke Jerry 📧 Email: chukwuka.jerry@gmail.com 📱 WhatsApp: +2348038782912 🔗 LinkedIn: linkedin.com 🐦 X (Twitter): @Mazimum_ 🏁 Future Work 📦 Add support for batch predictions via file upload 🌍 Extend model to …