Data-driven Parking Decisions: Proposal of Parking Availability Prediction Model

Due to the increase of the population and car ownership level, car-related problems which represented by traffic congestion and air pollution are caused. Parking problem is one of the most significant issues of them. Especially, long cruising time for parking spaces causes enormous economic cost. Thanks to recent IoT (Internet of Things) technology, it becomes possible to monitor availability of parking lots and some cities are providing monitored information to drivers. This system makes drivers to find parking spots easier than before. Nevertheless, there is a controversy over quality and quantity of availability information. Because of update interval and network delay, provided information is different from the real-time status (quality issues). Plus, there are still many unmonitored parking lots of which availability information couldn’t be read (quantity issues). In this paper, we propose two prediction models for qualitative and quantitative improvement of parking availability information. The proposed solution is evaluated using one month of occupancy rate data, publicly available from Seattle Department of Transportation.