Project Details

Project Title: Machine Learning Algorithm for Weather Forecasting

This project aimed to develop and evaluate machine learning models for predicting rainfall using weather data from the Ghana Meteorological Agency. The dataset included variables such as temperature (minimum and maximum), relative humidity, sunshine hours, wind direction, and wind speed. The study compared several models to identify the most accurate method for rainfall prediction. The Key objectives are to develop machine learning models to predict rainfall amounts, compare the performance of different models and to select the best-performing model for future deployment.

Methods and Algorithms:

K-Nearest Neighbors (KNN) Regression Random Forest (RF) Support Vector Regression (SVR) Decision Tree (DT) Data Preprocessing: Applied min-max normalization to scale features. Training and Testing: The dataset was split into 80% training and 20% testing.

Results

Evaluation Metrics: Root Mean Square Error (RMSE), Mean Square Error (MSE), Mean Absolute Error (MAE). Model Performance:
Random Forest: RMSE = 0.41 (best performance) Support Vector Regression: RMSE = 0.44 KNN Regression: RMSE = 0.42
Decision Tree: RMSE = 0.59

Conclusion

The Random Forest model outperformed other methods due to its robustness in handling complex data relationships. The study demonstrated that machine learning models, particularly ensemble methods like Random Forest, can significantly enhance rainfall prediction accuracy in Kumasi.