Prediction of COVID-19 pandemic measuring criteria using support vector machine, prophet and linear regression models in Indian scenario

Abstract

The whole world is embroiling the pandemic situation caused by COVID-19, which is spreading across all countries. As of mid-May, COVID-19 continues to increase the number of people affected and the number of deaths in each country. Each country's administrations concerned are making endless efforts to maintain public health, mental health and to regulate the rate of illness of COVID-19. Analysis of COVID-19 data using the machine learning paradigm is becoming a major interest of the researcher in these situations. Several researchers analyzed data from COVID-19 to predict infection, death, cured persons in the future, which may lead to the planning of each country's regulatory authority to maintain the public health of its people. The machine learning algorithm provides more accurate results when the data size is large due to the lower number of data sets available to COVID-19, making the most accurate predictions a challenging task to implement the machine learning algorithm. This paper was essentially designed to predict the active rate, the death rate, and the cured rate in India by analyzing the data of COVID-19. There are three models of machine learning Support Vector Machine (SVM), Prophet Forecasting Model, and Linear Regression Model for predicting active rate, death rate and cured rate. Prophet Forecasting Model has been shown to be the best predictive method for predicting active rate, death rate and cured rate compared to SVM and Linear Regression when the vast uncertain and small data sets.

Deep learning model for forecasting COVID-19 outbreak in Egypt

“…Finally, it suggested integrating machine learning and susceptible-exposed-infectious-removed (SEIR) models to enhance the accuracy and lead time of standard epidemiological models. Gupta et al (2020) employed a support vector machine (SVM), prophet forecasting model, and linear regression model for predicting the active, death, and cured rates in India. The prophet forecasting model produced better results compared to SVM and linear regression models because of the following points: 1) it analyzed the dataset trend to determine the growth curve, 2) it used the Fourier transformation series for data analysis based on daily, weekly, or yearly basis, and 3) it automatically identified the change points in data unlike the SVM and linear regression models.…”

“…Furthermore, the total number of active cases ( ), active rate ( ), death rate ( ), and recovered rate ( ) could be calculated using Equations (3) , (4) , (5) , (6) , respectively ( Gupta et al, 2020 ). Fig.…”

Prediction of the effective reproduction number of COVID-19 in Greece. A machine learning approach using Google mobility data

“…SVMs are based on the principle of data separation via a hyperplane that optimises data division, and has been suggested for COVID-19 modelling purposes (Gupta, Singh et al 2021). The regression tasks is made by estimating data points close to the hyperplane (support vectors) and minimizing the distance between those datapoints to a selected threshold epsilon ( E ), corresponding to error of tolerance.…”

Prediction of the effective reproduction number of COVID-19 in Greece. A machine learning approach using Google mobility data

“…SVMs are based on the principle of data separation via a hyperplane that optimises data division, and has been suggested for COVID-19 modelling purposes [54]. The regression tasks is made by estimating data points close to the hyperplane (support vectors) and minimizing the distance between those datapoints to a selected threshold epsilon ( , corresponding to error of tolerance.…”