hobbybwanali/Rainfall-Prediction-using-ML--Undergrad-Project-

Comparison of Logistic Regression, Random Forest, and MLP models for localized rainfall prediction using real data collected in Mkwinda, Malawi. # Rainfall Prediction Using Machine Learning – Mkwinda, Malawi This project compares the performance of three ML models for predicting rainfall occurrence: - Logistic Regression. - Random Forest - Multi-Layer Perceptron (MLP) The study used historical and real-time data collected using **Arduino-based tipping bucket rain gauges** that I designed and deployed. ## Features - Full preprocessing pipeline (KNN imputation, scaling, feature engineering) - Class imbalance handling (Random Oversampling) - Model evaluation with Accuracy, Precision, Recall, F1, ROC-AUC - January rainfall prediction for real-world validation. ## Results - **MLP achieved highest accuracy (83.6%)** - **Logistic Regression achieved highest recall (90%)** - **MLP had best balance of accuracy, precision & ROC-AUC** ## Tools Used Python, Pandas, Scikit-learn, Matplotlib, NumPy