A Novel Method for Prediction and Analysis of COVID 19 Transmission Using Machine Learning Based Time Series Models

• Published in: Wireless personal communications Vol. 133; no. 3; pp. 1935 - 1961
• Main Authors: Mann, Suman, Yadav, Deepshikha, Muthusamy, Suresh, Rathee, Dhruv, Mishra, Om Prava
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.12.2023; Springer Nature B.V
• Subjects: Communications Engineering; Computer Communication Networks; Coronaviruses; COVID-19; Deep learning; Disease transmission; Engineering; Growth factors; Influenza; Machine learning; Networks; Neural networks; Signal,Image and Speech Processing; Time series; Coronavirus; Time-Series; XBNet; Sarimax; Analysis; Prediction
• DOI: 10.1007/s11277-023-10836-z
• ISSN: 0929-6212

Coronavirus has been avowed world epidemic by the Organisation Mondiale de la Santé on March 11th 2020. Formerly, numerous investigators have endeavoured to envisage divergent periods of covid-19 malady and their possessions. Several have contemplate the temporal order of the events as primary factor which will contribute to the onset of infectious ailment including flu, influenza, etc. During this research analysis, total daily corroborated infested cases basedprognostication models for the time-series database of India for 30 days are estimated by applying extremely boosted neural network (XBNet). The main objective to introduce this XBNet model is to build an efficient and accurate time series deep learning model. Further, the performance of this model with previously developed time series models including logistic regression, facebook prophet, and sarimax is contrasted. To compare performance, numerous performance parameters like MAPE, RMSE, MAE, and MSE are employed to examine the consequence of model-fitting. This research also shows analysis of coronavirus cases based on three factors namely mortality rate, discharge rate, and the growth factor during different phases of lockdown. Also, projected the prediction of the cumulative number of confirmed COVID-19 cases for various time periods. This work presented the forecasts by applying the dataset that was attainable upto August 11th, 2021. The XBNet model showed a 99.27 percent precision accuracy and relatively less MSE, MAPE, RMSE, and MAE than other models. The results confirm superiority of the proposed approach over prevailing approaches.