Performance Improvement of Traffic Flow Prediction Model using Combination of Support Vector Machine and Rough Set

Deshpande, Minal; Bajaj, Preeti · 2017 · Crossref

DOI: 10.5120/ijca2017913473

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of short-term traffic flow prediction, a critical component of Intelligent Transportation Systems (ITS) that enables dynamic route guidance and traffic management. The authors identify that traffic flow is influenced by nonlinear and uncertain factors, making accurate prediction difficult. While Support Vector Machines (SVM) are recognized for their ability to approximate nonlinear systems with high accuracy, the study proposes a novel hybrid approach to further enhance performance. Specifically, the research investigates the use of Rough Set Theory (RST) as a post-processing validation tool to correct SVM predictions, aiming to reduce errors and improve the reliability of the forecasted traffic counts. The experimental design utilizes traffic data collected over six consecutive days (Monday to Saturday) in April 2014 at a location near the Perungudi toll plaza in Chennai, India. Data was gathered using The Infrared Traffic Logger (TIRTL), recording vehicle counts, speed, and other metrics. The authors prioritized traffic flow count for prediction, preprocessing the raw data by removing null and redundant entries and aggregating it into 5-minute, 10-minute, and 15-minute intervals. The SVM model employed a Gaussian Radial Basis Function (RBF) kernel, with a regularization parameter ($\gamma$) of 10 and a bandwidth ($\sigma^2$) of 0.2, selected through iterative testing. Data was normalized to a range of 0 to 1 to accelerate convergence. The Rough Set mechanism was applied to define lower and upper approximations of the target set based on the probability of occurrence in the training data. Predictions falling outside the lower approximation were corrected by replacing them with the nearest valid value from the training set. The results demonstrate that the hybrid model improves prediction accuracy compared to using SVM alone. Performance was evaluated using Mean Square Error (MSE). For a 5-minute aggregation interval using one past sample, the standard SVM model yielded an MSE of 1.42E-03. When the Rough Set validation was applied (SVM RS App), the MSE decreased to 1.38E-03. Visual analysis via scatter plots confirmed that while some deviations existed, the predicted values generally aligned closely with actual counts. The Rough Set method effectively handled outliers by ensuring predictions remained within the bounds of observed historical data, thereby minimizing error contributions from values not present in the training set. The study concludes that integrating Rough Set Theory as a post-processing validator significantly enhances the performance of SVM-based traffic flow prediction models. This approach provides a robust mechanism for handling the vagueness and nonlinearity inherent in traffic data, particularly in contexts like India where traffic patterns differ from Western settings. The findings suggest that this hybrid method offers a reliable solution for short-term prediction, supporting more effective real-time traffic control and management systems.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• telematics crash prediction