Comparative Analysis of Machine Learning Models for Predicting Heat Index Severity

Global warming has increased the frequency and intensity of extreme heat events, making heatwaves more prevalent and dangerous to human. These prolonged periods of severe heat can pose serious health hazards, including heatstroke, dehydration, and exhaustion, particularly in groups that are more susceptible to these conditions. The heat index, which combines air temperature and relative humidity, measures how hot it feels to the body and helps identify heat-related health risks. Accurate and timely prediction of heat index levels is essential for early warnings and public health preparedness. This study presents a comparative analysis of machine learning models for classifying the severity of heat index into five categories: 0 (safe), 1 (caution), 2 (extreme caution), 3 (danger), and 4 (extreme danger). Weather data from the Philippine Atmospheric, Geophysical, and Astronomical Services Administration (PAGASA) covering 2011 to 2022 were used to train and test different machine learning models, including Gaussian Naive Bayes (NB), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), and Logistic Regression (LR). Key parameters include temperature, humidity, wind speed, and rainfall, with temperature and humidity being the most significant factors in determining heat index levels. Performance metrics such as accuracy, precision, recall, and F1-score were used to evaluate each model. The results demonstrate that the Logistic Regression achieved the highest predictive performance, indicating its effectiveness in predicting the severity of the heat index.