Machine learning models for Forecasting Confirmed, Recovered and Deceased COVID-19 Cases in India

Abstract

Almost in every century some types of epidemic or pandemic diseases are impacting human lives and the economic system of the countries. Recently we came to know that a virus named Novel Corona virus spreads the disease COVID-19 to almost all over the world. It is very important to educate people about these types of pandemic diseases and its impact on human lives. It motivates us in doing this research to educate the people by forecasting spread of COVID-19 and how exponentially day-by-day the number of cases increased if not followed preliminary measures. The dataset named “cases_time_series” was collected from Kaggle data science community. This dataset represents statistics related to Indian people who effected due to COVID-19 disease. The dataset was extended by considering the actual values from authenticated sources. In the proposed work, experimentation has been done using three base learner models. 1. Linear Regression model 2. SMOReg model and 3. Gaussian Processes model. Forecasting done on three attributes using three learner models. Further the results were compared with actual values. From this comparison it is observed that the three modelsforecast the COVID-19 cases in a better way. If the nation does not follow the lock down rules and social distancing, numerous people may affect due to this disease.