Statistical analysis of correlation between weather parameters and new COVID-19 cases: a case study of Bosnia and Herzegovina

Abstract

This study analyzes the correlation between meteorological parameters and COVID-19 new daily cases for the period March-December 2020 in three large cities in Bosnia and Herzegovina: Sarajevo, Banja Luka and Mostar. The following weather parameters were included in the analysis: the mean temperature (TEMP), the maximum temperature (MAX), the minimum temperature (MIN), the mean wind speed (WDSP), the maximum sustained wind speed (MXSPD), the mean dew point (DEWP) and the mean visibility (VISIB).Two statistical approaches were performed in this study: Spearman rank correlation test and multiple linear regression method. Spearman rank correlation test indicated that there is a significant negative correlation between the number of new COVID-19 cases and TEMP, WDSP, MXSPD, MAX and VISIB in Banja Luka, TEMP, WDSP, MXSPD and VISIB in Sarajevo, while in Mostar all seven parameters are negatively correlated with the number of COVID-19 cases. In order to construct linear regression models, recursive feature elimination method was used for identifying independent variables. For each city, the proper linear regression models is constructed, with relatively satisfying values of R-squared value. To the best of our knowledge, no such study had been conducted in Bosnia and Herzegovina so far.