Trajectory of COVID-19 Data in India: Investigation and Project Using Artificial Neural Network, Fuzzy Time Series and ARIMA Models
Pradeep Mishra
Department of Statistics, College of Agriculture, Powarkheda, JNKVV (M.P.), India.
Chellai Fatih
Department of Based Education, University of Ferhat Abbas, Algeria.
Deepa Rawat
College of Forestry, Ranichauri, Tehri Garhwal, VCSG UUHF (Uttarakhand), India.
Saswati Sahu
West Bengal State University, W.B., India.
Sagar Anand Pandey
KVK, Bhatapara, IGKVV, Raipur, Chhattisgarh, India.
M. Ray
[RRTTS] (OUAT), Keonjhar, Odisha, 758002, India.
Anurag Dubey
Laboratoire de Mécanique Gabriel Lamé (LaMé), INSA Centre Val de Loire, Blois, France.
Olawale Monsur Sanusi
Laboratoire de Mécanique Gabriel Lamé (LaMé), INSA Centre Val de Loire, Blois, France.
*Author to whom correspondence should be addressed.
Abstract
Due to the impact of Corona virus (COVID-19) pandemic that exists today, all countries, national and international organizations are in a continuous effort to find efficient and accurate statistical models for forecasting the future pattern of COVID infection. Accurate forecasting should help governments to take decisive decisions to master the pandemic spread. In this article, we explored the COVID-19 database of India between 17th March to 1st July 2020, then we estimated two nonlinear time series models: Artificial Neural Network (ANN) and Fuzzy Time Series (FTS) by comparing them with ARIMA model. In terms of model adequacy, the FTS model out performs the ANN for the new cases and new deaths time series in India. We observed a short-term virus spread trend according to three forecasting models.Such findings help in more efficient preparation for the Indian health system.
Keywords: Artificial neural network, ARIMA, COVID-19 forecasting, fuzzy time series, India