On the Predictability of Parking Preferences

The recent advances in vehicular crowd sensing technology have paved the way for city-scale monitoring of parking spot availability. It enables intelligent parking navigation with the spot-level granularity, guiding drivers to the most convenient available spot within a busy parking lot. Since people would have various preferences on parking spot selection, personalization is a key to improve user experience in such parking navigation systems. In this paper, we investigate the feasibility of predicting a parking spot that a user would prefer out of a given set of available candidates.