Impact of COVID-19 in India and Its Metro Cities: A Statistical Approach

Abstract

The infectious coronavirus disease is spreading at an alarming rate, not only in India but also globally too. The impact of coronavirus disease (COVID- 19) outbreak needs to be analyzed statistically and modelled to know its behaviour so as to predict the same for future. An exhaustive statistical analysis of the data available for the spread of this infection, specifically on the number of positive cases, active cases, death cases and recovered cases, and connection between them could probably suggest some key factors. This has been achieved in this paper by analyzing these four dominant cases. This helped to know the relationship between the current and the past cases. Hence, in this paper, an approach of statistical analysis of COVID-19 data specific to metropolitan cities of India is done. A regression model has been developed for prediction of active cases with 10 lag days in four metropolitan cities of India. The data used for developing the model is considered from 26th April to 31st July (97 days), tested for the month of August. Further, an Artificial Neural Network (ANN) model using back propagation algorithm for active cases for all India and Bangalore has been developed to see the comparison between the two models. This is different from the other existing ANN models as it uses the lag relationships to predict the future scenario. In this case, data is divided into training, validation and testing sets. Model is developed on the training sets and is checked on the validation set, tested on the remaining, and then, it is implemented for prediction.