Spatiotemporal association between weather and Covid-19 explored by machine learning

The Covid-19 epidemic led to loss of the lives of many people in the world and had a very negative impact on the mental and physical health of humans. One of the effective ways to preventive strategies regarding is to study the impact of climatic parameters. This research introduces a new spatiotemporal methodology to explore the association between Covid-19 and hourly data of weather. This methodology developed based on machine learning using unsupervised clustering method. Six counties considered for finding association and the cities that have similar climatic temporal changes clustered and compared with cities that have similar number of Covid-19 cases. For this goal, a new model is developed for finding similarities between clusters, which indicates the association between weather and Covid-19. The result shows similarities are about 57% for wind speed, 63% for temperature, 63% for surface pressure, and 42% for elevation. Then result evaluated sing Kendall’s tau_b and Spearman’s rho which shows the proposed methodology has an acceptable result.