Data Driven Charging Station Placement.

With the rapid increasing availability of EV (electric vehicle) users, the demand for charging stations has also become vast. In the meanwhile, where to place the stations and what factors have major influence, remains unclear. These problems are bothering when EV companies tries to decide the locations for charging stations. Therefore, we tried to find an effective and interpretable approach to place them in more efficient locations. In common sense, a better location to place a station should relatively has a higher usage rate. Intuitively, we decided to predict usage rates of the candidate locations and tried to explain the result in the meantime, i.e. to find out how much important each feature is or what kind of influence they have. In this paper, we implement 2 models for the usage rate prediction. We also conduced experiments on real datasets, which contains the real charging records of anyo charging company in Shanghai. Further analysis is conducted as well for interpretation of the experiment result, including feature importance.