SEIR and Regression Model based COVID-19 outbreak predictions in India

Rajan Gupta; Gaurav Pandey; Poonam Chaudhary; Saibal Kumar Pal

In this study, two machine learning models SEIR and Regression were used to analyse and predict the change in spread of COVID-19 disease. We have analysed the data and found out that the number of cases per million in India is less than 0.5 till 30 th March 2020. Then, with the help of the SEIR model the value of R0 was computed to be 2.02. Also, we predicted the number of confirmed cases of COVID-19 for the next 14 days starting from 31 st March 2020 -13 th April 2020. During performance evaluation, our model computed the value of RMSLE for the SEIR model to be 1.52 and 1.75 for the regression model. Also, the value of spread of disease of R 0 was found to be 2.02. The result obtained from this study is taken from training data up to 30 th March 2020. Further, looking at the trend, there is definitely going to be an increase in the number of cases. Doctors, health workers and people involved in providing essential services have to be protected in accordance with prescribed medical norms. Community spreading in the future due to carelessness of individuals as well as groups can exponentially increase the number of cases. The peak is yet to come, hence the Government has to be extra vigil and enforce strict measures. In addition, provision of medical facilities across the country has to be aggressively enhanced. In future, an automated algorithm can be developed to fetch data in regular intervals and automatically predict the number of cases for weekly and biweekly data. In this way, Government and hospital facilities can also maintain a check on the supply and medical assistance / isolation required for new patients.