Predicting food consumer and customer behavior

Content Partner: Lincoln University. The aim of our research was to examine the comparability of food consumption patterns. Using the Random Forest method, we conducted pairwise comparisons on a representative sample of food customers from 6 countries. To carry out the prediction, factor and cluster analysis were performed and three clusters ("simplicity seekers", "demanding" and "practical housewives") were distinguished. In the analysis, the algorithm was applied to food shopper respondents in one country and the accuracy of the algorithm was tested in the other countries. Our results also further clarify the differences and similarities in food consumer behaviour across the 6 countries. Our results show that the Danish food consumption pattern is the most predictable. In particular, the responses of British, American and New Zealand respondents are the most suitable. It can also be seen from our results that geographical proximity does not necessarily give a country sample a good predictive ability. Finally, we have shown the homogenising effect of technology and social media use on food consumption and purchasing behaviour.