Evaluation of outliers detection algorithms for traffic congestion assessment in smart city traffic data from vehicle sensors

On-board sensors in vehicles are able to capture real-time data representations of variables conditioning the traffic flow. Extracting knowledge by combining data from different vehicles, together with machine learning algorithms, will help both to optimise transportation systems and to maximise the drivers' and passengers' comfort. This paper provides a summary of the most common multivariate outlier detection methods and applies them to data captured from sensor vehicles with the aim to find and identify different abnormal driving conditions like traffic jams. Outlier detection represents an important task in discovering useful and valuable information, as has been proven in numerous researches. This study is based on the combination of outlier detection mechanisms together with data classification methods. The output of the outlier detection phase will then be fed into several classifiers, which have been implemented to assess if the multivariate outliers correspond with traffic congestion situations or not.The research leading to these results has received funding from the ‘1HERMESSMART DRIVER’ project TIN2013-46801-C4-2-R (MINECO), funded by the Spanish Agencia Estatal de Investigación (AEI), and the ‘ANALYTICS USING SENSOR DATA FOR FLATCITY’ project TIN2016-77158-C4-1-R (MINECO/ERDF, EU) funded by the Spanish Agencia Estatal de Investigación (AEI) and the European Regional Development Fund (ERDF). And the first author was supported by the MINECO Grant BES-2014- 070462