Covid-19 cases prediction using SARIMAX Model by tuning hyperparameter through grid search cross-validation approach.

Area of Science:

• Epidemiology
• Data Science
• Public Health

Background:

• The COVID-19 pandemic, caused by SARS-Coronavirus, has had a significant global impact.
• Accurate prediction models are crucial for informing public health policy and decision-making.
• Existing models often lack the scope for accurate forecasting, necessitating improved predictive approaches.

Purpose of the Study:

• To forecast COVID-19 confirmed, recovered, and deceased cases in India for the next 20 days.
• To evaluate the performance of the SARIMAX model with hyperparameter tuning against other forecasting methods.

Main Methods:

• Utilized a dataset of COVID-19 cases from Kaggle.
• Employed the SARIMAX (Seasonal Autoregressive Integrated Moving Average with Exogenous Regressors) model.
• Optimized model hyperparameters using grid search cross-validation and Monte Carlo simulation for accuracy validation.

Main Results:

• The SARIMAX model with optimized hyperparameters achieved an R² score of 0.5112 and an RMSE of 1251.
• Forecasts indicate an increasing trend for confirmed and deceased COVID-19 cases in India.
• A decreasing trend was predicted for recovered COVID-19 cases.

Conclusions:

• The optimized SARIMAX model provides a reliable method for short-term COVID-19 case prediction in India.
• Findings highlight the need for continued monitoring and adaptive strategies to manage the pandemic.
• The study underscores the utility of advanced time-series forecasting techniques in public health.