Key points are not available for this paper at this time.
With increasing popularity of ride-hailing services, it becomes important to build transparent and explainable pricing models using artificial intelligence (AI). While the literature on this domain is growing steadily, the application of AI in pricing prediction is relatively new. We drew upon the New York City Taxi dataset to build pricing prediction models to bridge this gap. Our contributions are as follows. First, we created unique clusters for yellow and app-based cabs, leading to a dynamic pricing mechanism across different zones in New York City. Second, we converted a prediction problem into a classification problem by transforming the prices into four distinct quartiles. Third, we applied variable importance schemes to generate top predictors in each cluster. Fourth, our study reveals that differential effects of each predictor for cab-pricing across different clusters exist. Fifth, the "congestion surcharge" is significant for only a few clusters, and imposing such surcharges could hurt the overall taxicab industry. In this manner, our study contributes to the academic literature on taxicab pricing by offering transparent and actionable insights for stakeholders and policymakers, informed by robust AI-driven pricing models and empirical analyses of real-world data.
Rathore et al. (Thu,) studied this question.