Layered ensemble model for short-term traffic flow forecasting with outlier detection

Abstract

Real time traffic flow forecasting is a necessary requirement for traffic management in order to be able to evaluate the effects of different available strategies or policies. This paper focuses on short-term traffic flow forecasting by taking into consideration both spatial (road links) and temporal (lag or past traffic flow values) information. We propose a Layered Ensemble Model (LEM) which combines Artificial Neural Networks and Graded Possibilistic Clustering obtaining an accurate forecast of the traffic flow rates with outlier detection. Experimentation has been carried out on two different data sets. The former was obtained from real UK motorway and the later was obtained from simulated traffic flow on a street network in Genoa (Italy). The proposed LEM model for short-term traffic forecasting provides promising results and given the ability for outlier detection, accuracy, robustness of the proposed approach, it can be fruitful integrated in traffic flow management systems

Graded Possibilistic Clustering of Non-stationary Data Streams

“…Synthetic data sets containing concept drift (we select the Gaussian and electricity data sets) were generated using the Matlab program ConceptDriftData.m 4 [19]. We also Integrated our model in a traffic flow management system [20].…”

SVM‐based hybrid approach for corridor‐level travel‐time estimation

“…An online boosting approach for traffic flow forecasting under abnormal conditions was developed by Wu et al [21]. A layered ensemble model for short‐term traffic flow forecasting was proposed by Abdullatif et al [22]. Hamner [23] applied the random forest algorithm to predict travel‐time and showed that the proposed model outperformed other models in terms of estimation precision.…”

Tracking Time Evolving Data Streams for Short-Term Traffic Forecasting

“…1 An urban area of the city of Genoa, a town in the north-west of Italy, was mapped with the aid of Open Street Map data available at https://www.openstreetmap.org. Traffic parameters were obtained from actual observations and several days of traffic were simulated by using the SUMO open source traffic simulator [32].…”

“…We focus on the online approach to track and adapt to concept drift and shift and on using this knowledge to improve the ensemble forecasting model that was proposed in [1] by making the model able to not only detect outliers, but also track the changes in data streams.…”