Predicting Food POI Attractions for Smart Business using Passenger Commuting Patterns

Understanding client's choices and attractions to visit a place is always a valuable research for smart business recommendations. Food Point of Interests (FPOI) are topnotch places for revenue generation as thousands of people visit them every day. The biggest challenge is to know what attracts the visitors to commute to a particular FPOI. The understanding of demand with commuting pattern reveals important facts about the relationship of clients and the specific food type. In this paper, we analyzed the data of public transportation to find the preferences of public transport passengers attracted to FPOI. Perusing to forecast the attraction behavior, we have proposed a Food Choice Model based on Graph Neural Network. The aim is to develop a relative graph network between passenger's travel pattern and the FPOIs. Commuting patterns are extracted using Automated Fare Collection (AFC) records of Beijing Subway to ascertain the most traveled stations neighboring FPOIs. Our contribution is validated by FPOIs ranks and reviews data available socially. In the end, the results confirmed that our model is predicting the values in agreement with the social media data.