Method to Organize and Analyze Speed Data to Detect Unusual Traffic Activity

In this paper, we analyze weekdays' speed data from heavy traffic locations in Mexico City. The objective is to determine the "common denominator" speed from the weeks under study. For a specific location and time, speed data is compared between the same day for different weeks. Both a method and an algorithm are proposed to achieve this goal. The method conducts the following tasks: 1) organizes speed data of vehicles into "cell arrays", 2) estimates "usual traffic" values from the raw data through a heuristic technique, and 3) detects unusual speed values by a comparison of raw speed values against estimated usual traffic values. The algorithm is employed to determine the expected speed, and a metric score compares the expected and the observed values, which in turn returns the locations and times that present anomalous results. The twelve top locations with the worst traffic conditions in Mexico City during 2016 were chosen. Measurements were realized in 2018. Results show that it is possible to detect the locations, days and times when the vehicles' speed was unusual. This indicates that it is desirable to conduct further investigations to determine the causes inducing significant deviation, higher or lower, from the expected speed.