A Reproducible Pipeline for Leveraging Operational Data through Machine Learning in Digitally Emerging Urban Bus Fleets

This dataset supports the study "A Reproducible Pipeline for Leveraging Operational Data through Machine Learning in Digitally Emerging Urban Bus Fleets" and contains operational data from a fleet of 164 hybrid diesel buses. The data were collected from the vehicles' onboard diagnostics (OBDII) and telematics systems, with a particular focus on variables related to the Diesel Particulate Filter (DPF) behavior. The goal is to develop and evaluate predictive models capable of estimating the duration a vehicle remains in critical soot accumulation zones (Zones 3 and 4), in order to support predictive maintenance decisions.

Two files are included:

First batch.xlsx – Contains the preprocessed data used to train and validate the initial regression models. It includes input features such as time, distance, and fuel since last regeneration, as well as the soot zone label and the target variable (duration in zone).

Second batch.xlsx – Contains a separate one-month batch of unseen data used to simulate model deployment and assess performance under real-world conditions.

All variables have been anonymized to preserve confidentiality while maintaining the structure required for temporal analysis.