Restaurant health inspections aim at identifying health violations and shall reduce the risk that restaurant visitors suffer from foodborne illness. Nevertheless, regulatory authorities’ resources are limited, so an efficient mechanism that supports scheduling of health inspections is necessary.
We build upon information efficiency theory and investigate whether information extracted from online review platforms is useful to predict restaurant health violations. Furthermore, we examine how the expectation disconfirmation bias impacts classification performance. Analyzing a large sample of health inspections, corresponding online reviews and restaurant visitor data, we propose and evaluate different predictive models.
We find that classifiers specifically taking into account information from online review platforms outperform different baseline approaches. We thus show that online reviews encompass private information indicating strong information efficiency. Furthermore, we observe that the expectation disconfirmation bias has an influence on classification performance in case of restaurants with a low star rating and with a poor inspection history. An ensemble classifier can help to mitigate this influence. Thus, online review platforms contain relevant information to predict future health violations. Our results are highly relevant for regulatory authorities, restaurant visitors and restaurant owners.
Leveraging online review platforms to support public policy; Predicting restaurant health violations based on online reviews
Decision Support Systems