Changing shopping habits in supermarket during Covid-19

Official URL: https://risc01.sabanciuniv.edu/record=b2579239_(Table of contents)

On December 31, 2019, the world was shaken by the news of a new virus observed in Wuhan city, Hubei, China. Covid-19 affected all the industries but mostly the retail industry. Stock-pilling and panic buying behaviors were seen in supermarkets. Most of the retailers could not satisfied the basic demands with their supply plans. Therefore, the main aim of this research is to find out the changing habits and consumer behaviors due to the effect of COVID-19. The analysis was conducted using the CRISP-DM model steps. The data set was revealed from one of the large retailers in Turkey. Initially, cluster analysis was conducted to group the products according to their similarity in terms of shopping behaviors. Subsequently, stepwise regression, best subset selection and Lasso models were applied to each cluster in order to analyze the relation between consumer shopping behavior and Covid cases. The best regression model for each cluster is selected using 10-fold cross-validation as well as Cp and adjusted R2. These analyses were applied both to the physical and online market data to highlight the shift from physical to online market. Eventually, this research found out that there is a significant shift from physical to online market during Covid-19. Many of the product groups have seen a significant increase in sales compared to 2019 with the effect of panic-buying. The shifts in sales levels are statistically related with the Covid-19 case numbers and curfews.