Using SCATS data to predict bus travel time

Abstract

The provision of accurate travel time information of public transport vehicles is valuable for both operators and passengers. It helps operators effectively implement their management strategies. It also allows passengers to schedule their departure to minimize waiting times. Public transport travel time is affected by several factors such as traffic flow, passenger demand, etc, which have to be considered to make precise predictions. However, previous studies have not explicitly considered real world traffic flow variables in their prediction models. This paper aims at using traffic flow data to predict bus travel time, and at examining the value which traffic flow data could make to the accuracy of predictions. It uses travel time values obtained from GPS recorded data from a bus route in Melbourne, Australia, to develop three models. The first model is an artificial neural network that uses saturation degree data collected by the Sydney Coordinated Adaptive Traffic System (SCATS) at intermediate signalized intersections along with schedule adherence to predict bus travel time. A historical data based neural network that uses temporal variables (time of day, day of week and month of year) and schedule adherence as well as a model that traditionally utilizes scheduled travel times for future travel time predictions are also developed. The results show that the first model outperforms other models.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 9 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.