TEPs uses traffic count data to predict periodic and annual volumes at the level of each individual roadway. TEPs-I is a chain of models for traffic volume and consists of six sub-models including novel spatio-temporal aggregation and disaggregation models, data pattern recognition, advanced interpolation, and machine learning methods. Models at the top of chain address stations that have long records of data, and feed information to lower level models.
A key for better prediction. TEPs feeds information from permanent monitoring sites to temporary sites for better traffic prediction
TEPs uses a network-based interpolation technique to predict traffic volume for all roads, even for locations without monitoring sites.
TEPs can be used at any location and scale with minimum data requirement.
TEPs generates maps of traffic volumes across an urban road network
TEPs has a very user friendly interface.
TEPs can predict 10 years of traffic volumes for all roads in a metropolitan area in a single day
PRTCS, KCOUNT and LocalSVR
TEPs-I is a chain of models for traffic volume prediction and consists of six sub-models including novel spatio-temporal disaggregation and aggregation sub-models (PECOUNT-I and PECOUNT-II), data pattern recognition model (PRTCS), advanced interpolation model (KCOUNTS), machine learning-based model (Local-SVR), and optimization model (OPTIM-STATIONS). Models at the top of chain feed information to lower level models.
TEPs-I was used to predict periodic and annual volumes at the level of each individual roadway (more than 50000 roads) from 2007 to 2016 in the City of Toronto. Outputs of sub-models (PECOUNT-I & II; PRTCS; KCOUNTS and Local-SVR) were validated against observed traffic counts, indicating robust model performance.
This project was supported by The City of Toronto's Transportation Services Division and Atmospheric Fund
Copyright © 2019 Traffic Emission Prediction scheme (TEPs) - All Rights Reserved.
Updated on Feb 28, 2019