Analysis on novel coronavirus (COVID-19) using machine learning methods

Abstract

In this paper, we are working on a pandemic of novel coronavirus (COVID-19). COVID-19 is an infectious disease, it creates severe damage in the lungs. COVID-19 causes illness in humans and has killed many people in the entire world. However, this virus is reported as a pandemic by the World Health Organization (WHO) and all countries are trying to control and lockdown all places. The main objective of this work is to solve the five different tasks such as I) Predicting the spread of coronavirus across regions. II) Analyzing the growth rates and the types of mitigation across countries. III) Predicting how the epidemic will end. IV) Analyzing the transmission rate of the virus. V) Correlating the coronavirus and weather conditions. The advantage of doing these tasks to minimize the virus spread by various mitigation, how well the mitigations are working, how many cases have been prevented by this mitigations, an idea about the number of patients that will recover from the infection with old medication, understand how much time will it take to for this pandemic to end, we will be able to understand and analyze how fast or slow the virus is spreading among regions and the infected patient to reduce the spread based clear understanding of the correlation between the spread and weather conditions. In this paper, we propose a novel Support Vector Regression method to analysis five different tasks related to novel coronavirus. In this work, instead of simple regression line we use the supported vectors also to get better classification accuracy. Our approach is evaluated and compared with other well-known regression models on standard available datasets. The promising results demonstrate its superiority in both efficiency and accuracy.

Keywords: Active cases; COVID-19; Novel coronavirus; Pearson; Polynomial regression; Recoveries; Simple linear regression; Support vector regression model.

Fig. 1

COVID-19 Statistics published by WHO . (A) Total Number of COVID-19 world-wise effected active cases (B)Closed cases (Death and Recovery rate) weekly wise (C) World-wide total number of COVID-19 cases (D) Total world-wide coronavirus deaths.

Fig. 2

SVM Model Maximum-margin Hyperplane.

Fig. 3

Accuracy of predicting total number of positive cases in different regions; (a) Mainland China, (b) US, (c) Italy, (d) South Korea, (e) India.

Fig. 4

Predicting total number of active cases in different regions; (a) Mainland China, (b) US, (c) Italy, (d) South Korea, (e) India.

Fig. 5

Predicting recoveries in different regions; (a) Mainland China, (b) US, (c) Italy, (d) South Korea, (e) India.

Fig. 6

Result Analysis of transmission of COVID-19 in (a) Mainland China, (b) US, (c) Italy, (d) South Korea, (e) India.

Fig. 7

Result Analysis of different correlation information of New York City (a) correlation table for temperature, humidity, wind speed, and total positive COVID19 cases, (b) correlation between temperature and total positive cases, (c) correlation between total positive cases and humidity.

Fig. 8

Result Analysis of different correlation information of Milan city (a) correlation table for temperature, humidity, wind speed, and total positive COVID19 cases, (b) correlation between humidity and wind Speed, (c) correlation between total positive cases and humidity, (d) correlation graph between total positive cases and wind speed.