A COMPARISON OF LOGISTIC REGRESSION AND GEOGRAPHICALLY WEIGHTED LOGISTIC REGRESSION (GWLR) ON COVID-19 DATA IN WEST SUMATRA
Irvanal Haq,Muhammad Nur Aidi,Anang Kurnia,Efriwati Efriwati
DOI: https://doi.org/10.30598/barekengvol17iss3pp1749-1760
2023-09-30
Abstract:An understanding of factors that affect the recovery time from a disease is important for the community, medical staff, and also the government. This research analyzed factors that affect the recovery time of Covid-19 sufferers in West Sumatra. In addition, the consumption of a herbal made from Sungkai leaves, which is believed by some people in West Sumatra to accelerate the healing from Covid-19, was also included in the analysis. The recovery time here was categorized into two classes (binary): 1 for within 2 weeks, and 0 for more than 2 weeks. The methods used were logistic regression and geographically weighted logistic regression (GWLR). GWLR provides estimates of parameters for each location. The data used in this study is Covid-19 data of 2021 taken from the Regional Research and Development Agency (Litbangda) of West Sumatra with a total of 764 observations collected from 19 regencies/cities in West Sumatra. The results showed that there was no difference between the logistic regression model and the GWLR models based on the values of AIC and the ratio of deviance and degrees of freedom (df). The addition of spatial factors through GWLR models did not provide additional information regarding the recovery of Covid-19 sufferers within 2 weeks or more than 2 weeks. The logistic regression model gives the result that, at significance level α = 10%, residence, vaccination status, and symptoms significantly affect the recovery time within 2 weeks or more for Covid-19 sufferers, while other variables, namely sex, age, Sungkai leaves consumption status, and ginger consumption status have no significant effects.