Autoencoder signals do not outperform classic methods detecting bus delays and anomalies
Detecting bus delays and service anomalies in Niterói with autoencoder latent feature representations
Machine Learning
Summary
Bus passengers care about trips that are late or days when bus service is generally bad. The authors studied if a type of AI model called an autoencoder, which compresses raw bus GPS data, could detect both late trips and problem days. They found that the autoencoder didn’t do better than traditional methods at spotting delays and that some signals it gave about days with issues were misleading. However, the model did consistently pick up some general day-level patterns, mostly related to the type of day and quality of GPS signals.
What this means in practice
- •For public transit planners: Use extracted daily patterns from GPS data to monitor general bus system reliability despite limited delay detection improvement.
- •For urban transport analysts: Employ autoencoder-based anomaly scores as a supplementary indicator for system-wide unusual days, validated beyond weather effects.
Tested on one dataset.
Authors
Matheus Germano, Claudio E. C. Campelo
Abstract
Bus riders judge service by two failures: single trips that run late, and days on which the whole system misbehaves. We ask whether one unsupervised representation of raw GPS telemetry can serve both diagnoses on the NetMob 2026 Niterói bus dataset. An autoencoder compresses each trip into a 64-dimensional latent vector; the frozen encoder feeds supervised delay classifiers, and the reconstruction error serves as an anomaly score. Two findings are negative. First, the latent representation does not improve delay detection: gradient boosting on per-trip summary features reaches $0.846 \pm 0.013$ ROC-AUC against $0.838 \pm 0.016$ for the best latent setup, and the gap becomes significant once the preprocessing filters are relaxed. Second, two of the three external checks of the anomaly score were circular. Retraining without rainfall removes the rain association ($5.4\%$ against $5.0\%$ flagged, $p = 0.84$) and an apparent citywide disruption on 23 March. What survives is a day-level signal stable across three autoencoders (Spearman $ρ\geq 0.96$), driven mostly by day type, with degraded-telemetry days still separable within weekday and weekend strata. The alarm threshold limits detection more than the score does: a pooled percentile cut flags $5.7\%$ of trips on the worst degraded day, while re-scaling within each bus line flags $29.3\%$. The delay model rests almost entirely on one feature, no latent dimension is individually interpretable, and 22 chronic line-directions carry half of all delayed trips.