Trajectory Data Mining and Trip Travel Time Prediction on Specific Roads

TL;DR

This study tackles travel time prediction for frequent road segments in Islamabad, Pakistan, by building a local trajectory-data pipeline and evaluating three neural approaches: a shallow ANN, a deep MLP, and an LSTM. By mining GPS trajectories, applying map matching to the OSM network, and simplifying trajectories, the authors construct a rich 12-feature dataset for modeling. Across six busy routes, the LSTM model delivers the lowest RMSE/MAE compared with ANN and MLP, underscoring the value of temporal modeling for urban travel-time forecasting. The work demonstrates the feasibility and practical impact of data-driven TTP in a developing-city context, with potential applications in navigation services and ITS deployments.

Abstract

Predicting a trip's travel time is essential for route planning and navigation applications. The majority of research is based on international data that does not apply to Pakistan's road conditions. We designed a complete pipeline for mining trajectories from sensors data. On this data, we employed state-of-the-art approaches, including a shallow artificial neural network, a deep multi-layered perceptron, and a long-short-term memory, to explore the issue of travel time prediction on frequent routes. The experimental results demonstrate an average prediction error ranging from 30 seconds to 1.2 minutes on trips lasting 10 minutes to 60 minutes on six most frequent routes in regions of Islamabad, Pakistan.

Figures (13)

• Figure 1: Map Matching
• Figure 2: 861 points reduced to 46 points only
• Figure 3: Complete Data Processing.
• Figure 4: Visualisation of overall data
• Figure 5: ANN used in experiments