Clustering in Before-After Studies of ITS Measures

Intelligent Transport System (ITS) measures are becoming more and more common on urban motorways. Different methods can be used to evaluate the effectiveness of the ITS measure, for example traffic simulation, cost-benefit analysis and before-after studies. A fundamental part of a before-after study is the choice of days for comparison of the before and after periods. Many previous before-after studies of ITS measures are based on simple methods for choosing comparable days, similar to the benchmark method used in this paper. For the chosen case study, a modified Variable Speed Limit (VSL) control strategy, we have used clustering to 1) quickly find similar days for evaluation, 2) identify type days with different characteristics and 3) identify deviating days that are non-representative for the before-after study. From the case study, it is clear that the choice of the days for evaluation are important and can generate different inference, and the clustering method provides promising results compared to the benchmark.