Description of an Electric Vehicle Charge Station Network using Knockoff Filters

Abstract

In this work we analyse the utilisation of Electric Vehicle (EV) public charging stations in the Netherlands to understand and describe their usage as a function of surrounding premises (such as hospitals, casinos and schools, among others) and population. Also, we analyse the charging performance of such charging stations taking into account temporal values and charging measures taken from transactions registered within the years 2012 and 2016. In order to identify the (potentially) explanatory variables that are meaningful, we will use a False Discovery Rate (FDR) control approach known as Knockoff filters. Results reveal that charging stations located close to Kindergartens, Fuel stations and Car sharing points are more likely to be used more frequently and for the longest time; whereas those users who charge their vehicles either on a weekend or in July between 12 AM and 6 AM are expected to charge their vehicles faster than in other configurations.