Creative Commons

One of the most critical functions of the modern Intelligent Transportation Systems (ITS) is the accurate and real - time short - term traffic prediction. This function becomes even more important under the presence of at ypical traffic conditions. In this disserta- tion , we propose a novel hybrid method for short - term traffic prediction under both typ- ical and atypical conditions. An Automatic Incident Detection (AID) algorithm that is based on Support Vector Ma- chines (SVM) is utilized to check for the presence of an atypical event (e.g. traffic acci- dent). If one occur s, the k - Nearest Neighbors (k - NN) non - parametric regression model is used to predict traffic. If no such case occurs, the Autoregressive Integrated Moving Avera ge (ARIMA) parametric model is activated. In order to evaluate the performance of the proposed model, we use open real world traffic data from the Caltrans Performance Measurement System (PeMS). We compare the proposed model with the unitary k - NN and ARIMA models. Preliminary results in- dicate that the proposed model outperforms its competitors in terms of prediction accu- racy under both typical and atypical traffic conditions.