Analyzing Taste Preferences From Crowdsourced Food Entries

Abstract

It is well known that in a balanced diet, eating the right amount of calories and nutrients to maintain a healthy weight is important for one's physical wellness and health. Thus, understanding demographic and behavior patterns of food consumption is a topic that several researchers in public health have long pursued. In this paper, we study how perceived food tastes, which are known to affect palatability of foods, are related to the dynamics and nature of population-wide dietary preferences and patterns over demographics, time, and location. In contrast to previous studies that have been clinical in nature based on small samples of participants via treatment data, our study offers a more ``big data''-style approach by leveraging a massive collection of food items and entries from the MyFitnessPal user base. Despite its differences from traditional research, our findings actually validate some previous studies that correlate food taste with certain population groups or public health patterns. In addition, we are able to extend research into previously unexploited directions.