Street Level Travel Time Estimation Employing Wbpst-svr Model Using Gtfs Feature

Abstract

Accurate Estimation Of Street-level Travel Time Is A Fundamental Requirement For Intelligent Transportation Systems (ITS), Enabling Effective Traffic Management, Route Optimization, And Real-time Traveler Information Services. With The Increasing Availability Of Urban Mobility Data, Particularly From Public Transport Systems, Data-driven Approaches Have Gained Prominence Over Traditional Rule-based Or Simulation-based Models. General Transit Feed Specification (GTFS) Data Provides A Standardized And Rich Source Of Spatio-temporal Information Related To Transit Schedules, Routes, Stops, And Frequencies, Making It Highly Suitable For Fine-grained Travel Time Estimation At The Street Level. This Paper Presents A Wavelet Tree And Support Vector Regression (SVR) Model For Forecasting Street Level Travel Time Employing GTFS Features. The WBPST Model Has Been Used For Filtration While The SVR Model Has Been Used For Pattern Recognition. The Results Show That The Proposed Model Attains An MAPE Of Just 2.44% At 68 Iterations. The Model Also Predicts The Level Of Congestion With An MAPE Of 2.14%. The Model When Compared With Existing Benchmark Models Can Be Seen To Improve Upon The Existing Results.

Keywords