Accuracy of open-data versus operator real-time bus arrival predictions: a paired evaluation of the Bide+ method in Biscay, Spain

This paper compares an operator's real-time bus predictions with an independent predictor for the same buses at the same moments, for Bizkaibus, the interurban network of Biscay (Spain), which publishes vehicle positions and timetable as open data and shows its own waiting times in an app. For one week in November 2026, the official waiting times of 60 stops were recorded every minute with the position feed. Each official announcement was matched to the bus it referred to, confirmed by the distance the service reports, and compared with the published timetable and with the Bide+ method, which predicts arrivals from the open data alone and was frozen before the week. All three were checked against the actual passage, reconstructed from the feed and from GPS positions. Across 555,744 paired predictions on six days, the timetable was within two minutes of the actual passage in 16% of cases, the official real time in 39% and Bide+ in 64%. Below three minutes the two real-time sources were equally often within two minutes, although the official one had the smaller mean error below six minutes; from three minutes on, Bide+ was more often within two minutes, by 29 to 35 percentage points at 10–20 minutes (95% interval by stop), at 57 of 59 stops and at every hour. An ablation attributes its advantage to learned segment travel times. The results apply to the 54% of first-bus announcements whose bus was tracked and confirmed, and to one week; data and code are open. Files: the preprint in journal format (PDF), a reading version with the same content (PDF), and an archive with the capture program, the comparison and analysis code, the result tables, the figure data, the timetable database and the learned tables used. Code under the MIT licence; text and derived data under CC BY 4.0. The raw captures (feed copies and official responses) are not included: they are assets of the release comparacion-2026-11 of the GitHub repository, and the official responses are not covered by the CC BY licence. Preprint, not peer reviewed.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24286026
Primary Topic
Transportation Planning and Optimization
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Accuracy of open-data versus operator real-time bus arrival predictions: a paired evaluation of the Bide+ method in Biscay, Spain

Yeray Garrido
2 links
Zenodo (CERN European Organization for Nuclear Research)
Transportation Planning and Optimization
preprint

Accuracy of open-data versus operator real-time bus arrival predictions: a paired evaluation of the Bide+ method in Biscay, Spain

Yeray Garrido
preprint en

Abstract

This paper compares an operator's real-time bus predictions with an independent predictor for the same buses at the same moments, for Bizkaibus, the interurban network of Biscay (Spain), which publishes vehicle positions and timetable as open data and shows its own waiting times in an app. For one week in November 2026, the official waiting times of 60 stops were recorded every minute with the position feed. Each official announcement was matched to the bus it referred to, confirmed by the distance the service reports, and compared with the published timetable and with the Bide+ method, which predicts arrivals from the open data alone and was frozen before the week. All three were checked against the actual passage, reconstructed from the feed and from GPS positions. Across 555,744 paired predictions on six days, the timetable was within two minutes of the actual passage in 16% of cases, the official real time in 39% and Bide+ in 64%. Below three minutes the two real-time sources were equally often within two minutes, although the official one had the smaller mean error below six minutes; from three minutes on, Bide+ was more often within two minutes, by 29 to 35 percentage points at 10–20 minutes (95% interval by stop), at 57 of 59 stops and at every hour. An ablation attributes its advantage to learned segment travel times. The results apply to the 54% of first-bus announcements whose bus was tracked and confirmed, and to one week; data and code are open. Files: the preprint in journal format (PDF), a reading version with the same content (PDF), and an archive with the capture program, the comparison and analysis code, the result tables, the figure data, the timetable database and the learned tables used. Code under the MIT licence; text and derived data under CC BY 4.0. The raw captures (feed copies and official responses) are not included: they are assets of the release comparacion-2026-11 of the GitHub repository, and the official responses are not covered by the CC BY licence. Preprint, not peer reviewed.

Zenodo (CERN European Organization for Nuclear Research)
Sustainable cities and communities, Industry, innovation and infrastructure
Transportation Planning and Optimization
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon