Forecasting of COVID-19 Cases in INDIA Using ARIMA and AR Time-Series Algorithm

ABSTRACT

The COVID-19 pandemic has been spreading and affecting worldwide. On 30th January 2020 India reported its first coronavirus confirmed case. The main aim of the proposed work is to devise an algorithm for prediction of Covid-19 cases in India. In this paper, we propose to use time-series algorithms, Autoregressive Integrated Moving Average (ARIMA) and Autoregressive (AR). We have simulated the designed algorithm with the dataset of COVID-19 till 20th February, 2021 for the wave1 and from 01st March, 2021 till 25th September, 2021 we collected data for wave2 and generated 6-days forecasts of confirmed, recovered and death cases. During the 1st wave we observed that there might be another wave 2, after analyzing the wave1 dataset, of coronavirus as result shows that Covid-19 confirmed cases are rising rapidly. Proposed research observations show that the death rate is decreasing, and recovery rate is increasing, one of the possible reasons is herd immunity and vaccination. We are comparing actual cases with forecasting coronavirus cases. ARIMA based models are showing promising results over AR based models. The most difficult part doing this work is to identify parameters due to sudden increase-decrease trend in coronavirus cases. The proposed work reports quality scoring metrics of forecasting for both the models. This will help future researchers to find the best outcome among Auto Regressive Integrated Moving Average (ARIMA) and Auto Regressive (AR) based models. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.