DanielOdushe/NGA-Weather-Analysis

Spatio-Temporal Analysis of Weather Patterns in Nigeria on 2/21/2024 # Spatio-Temporal Analysis of Weather Patterns in Nigeria on 2/21/2024 ## Project Overview This project analyzes weather patterns across Nigeria and builds machine learning models to predict temperature based on various meteorological and geographical features. The analysis covers 6 regions, 144 cities, and 22,541 data points, providing comprehensive insights into Nigeria's climate patterns. ## Dataset Characteristics - **Total Records**: 22,541 weather observations - **Geographical Coverage**: 6 regions (Anambra, Borno, Kaduna, Kogi, Lagos, Rivers) - **Features Analyzed**: Temperature, humidity, pressure, wind speed, cloud cover, population, and geographical coordinates - **Temperature Range**: 17.9°C to 41.0°C (Mean: 30.2°C) ## Key Insights ### Regional Weather Patterns - **Hottest Region**: Anambra (mean temp: 31.58°C) - **Coolest Region**: Kaduna (mean temp: 27.33°C) - **Most Humid**: Lagos (67.0% humidity) - **Driest**: Borno (21.89% humidity) - **Highest Pressure**: Kaduna (1011.91 hPa) - **Windiest**: Borno (3.94 m/s average wind speed) ### Feature Correlations with Temperature - **Strong Negative Correlation**: Pressure (-0.640) - **Moderate Negative Correlation**: Humidity (-0.296) - **Positive Correlation**: Wind Speed (0.221) ### Feature Importance (Random Forest) 1. **Pressure** (48.28%) - Most significant predictor 2. **Humidity** (25.21%) 3. **Latitude** (6.67%) 4. **Wind Speed** (6.50%) 5. **Cloud Cover** (5.75%) ## Machine Learning Models Performance ### Model Comparison | Model | MSE | RMSE | MAE | R² | Cross-Validated R² | |-------|-----|------|-----|----|-----------------------| | Random Forest | 0.202 | 0.449 | 0.223 | 0.987 | 0.986 | | Gradient Boosting | 1.681 | 1.297 | 0.957 | 0.893 | 0.891 | | Support Vector Regression | 1.839 | 1.356 | 0.903 | 0.883 | 0.884 | | Linear Regression | 6.018 | 2.453 | 2.005 | 0.617 | 0.634 | | Ridge Regression | 6.018 | 2.453 | 2.005 | 0.617 | 0.634 | | Lasso Regression | 6.154 | 2.48 …