Hbabaldu1/Khausa_Project

Final Year Sule Lamido University, Kafin Hausa, B.sc Physics Project on Kafin Hausa Weather Prediction Using Machine Learning # Khausa Weather Prediction System A machine learning-based weather prediction system developed as a B.Sc Physics final year project at Sule Lamido University, Kafin Hausa. This project uses historical weather data from Kafin Hausa to predict next-day temperatures using various machine learning regression models. ## Project Overview This project implements and compares multiple machine learning algorithms to forecast daily temperature in Kafin Hausa, Nigeria. The system processes NASA POWER daily weather data and trains three different regression models to determine which approach provides the most accurate predictions. ## Dataset - **Source**: NASA POWER (Prediction of Worldwide Energy Resources) - **Location**: Kafin Hausa, Nigeria (Latitude: 12.024°N, Longitude: 9.91°E) - **Time Period**: January 1, 2000 - January 1, 2026 - **Variables Used**: - T2M: Temperature at 2 meters - RH2M: Relative Humidity at 2 meters - PS: Surface Pressure - WS2M_MAX: Maximum Wind Speed at 2 meters ## Features ### Data Processing - Automated CSV data loading and cleaning - Conversion of Year + Day-of-Year to standard date format - Automated target variable creation (next-day temperature) - Missing value handling ### Machine Learning Models The project implements and evaluates three regression models: 1. **Linear Regression** - Baseline model for comparison - Currently commented out but available for testing 2. **Decision Tree Regressor** - Non-linear approach to capture complex patterns - Currently commented out but available for testing 3. **Random Forest Regressor** (Active Model) - Ensemble method with 100 trees - Best performance for temperature prediction - Currently selected as the primary prediction model ### Evaluation Metrics - **Mean Absolute Error (MAE)**: Average absolute difference between predicted and actual values - **Mean Squared Error (MSE)**: Squared differences to penalize larger errors ### Analysis & Visualization - Feature importance analysis showi …