Data Analysis and Forecasting of COVID-19 Outbreak in India Using ARIMA Model

Kaka, B.; Garg, D.; Goel, P.; Ganatra, A.

Nowadays COVID-19 has created a pandemic for the whole world. This is also known as Novel Coronavirus-2019. In this paper, time series analysis using the ARIMA model is brought forward for COVID-19 prediction on the confirmed cases in India. ARIMA model can significantly give precise forecast results based on AIC (Akaike Information Criteria) value. ARIMA model can considerable reduce the errors of the prediction results with 24418 AIC value for predicting confirmed cases in India. The work is implemented gathering the data of confirmed cases from different states of the country. The duration from 30th January 2020 to 28th April 2020 has been taken into consideration for verifying the positive cases of corona in India. Moving average and auto regressive models are used for accurate prediction and maintaining seasonal differencing and second order differencing. The graphical representation is demonstrated applying the technique named Data Visualization in python programming. It shows the increasing amount of confirmed cases as well as the number of cured cases and death cases in India. It is examined that the p, d, q parameter in ARIMA can locate the best AIC value. According to the analysis in this context rolling mean and standard deviation test Statistic value is −1.186895. ADF Statistic value is 1.186895. Data sets are divided in training and testing module respectively for approximate judgement of positive cases. © 2021, Springer Nature Singapore Pte Ltd.