This paper proposes an improved nonparametric regression (INPR) algorithm for forecasting traffic flows and its application in automatic detection of traffic incidents. The INPRA is constructed based on the searching method of nearest neighbors for a traffic-state vector and its main advantage lies in forecasting through possible trends of traffic flows, instead of just current traffic states, as commonly used in previous forecasting algorithms. Various simulation results have indicated the viability and effectiveness of the proposed new algorithm. Several performance tests have been conducted using actual traffic data sets and results demonstrate that INPRs average absolute forecast errors, average relative forecast errors, and average computing times are the smallest comparing with other forecasting algorithms.
Traffic-incident detection-algorithm based on nonparametric regression
IEEE Transactions on Intelligent Transportation Systems ; 6 , 1 ; 38-42
2005-03-01
637017 byte
Article (Journal)
Electronic Resource
English
Traffic-Incident Detection-Algorithm Based on Nonparametric Regression
British Library Conference Proceedings | 2005
|Application of nonparametric regression in predicting traffic incident duration
DOAJ | 2018
|TRAFFIC INCIDENT DETECTION ALGORITHM BASED ON NON-PARAMETER REGRESSION
British Library Conference Proceedings | 2002
|