Predicting daily new cases of COVID-19 using data mining techniques

Accurate forecasting of daily new COVID-19 cases remains a critical challenge for effective epidemiological surveillance and healthcare system preparedness. This study proposes a data-driven predictive framework that leverages advanced data mining techniques to model daily COVID-19 case trends using a comprehensive dataset comprising case statistics, demographic attributes, vaccination status, cluster information, and temporal indicators. The methodology involves systematic data preprocessing, feature engineering—including lagged and temporal variables—and the application of regression-based and time-series forecasting models. Model performance is rigorously evaluated using standard statistical error metrics to assess predictive reliability. The results indicate that incorporating vaccination categories, cluster-related features, and historical lag variables substantially enhances forecasting accuracy, enabling the models to capture nonlinear dynamics and temporal dependencies in case progression. The findings demonstrate the effectiveness of data mining approaches in improving short-term COVID-19 case prediction, thereby providing a technically robust framework for real-time epidemic modeling and informed public health decision-making.