CAD-Scania

Dataset Summary

CAD-Scania is a single-source continual anomaly detection benchmark scenario for predictive maintenance in the automotive/industrial domain. It is derived from the SCANIA Component X Dataset, a real-world multivariate time-series dataset of anonymized engine-component operational readouts, repair records, and specifications from SCANIA trucks, and converts it into a sequence of concept-grouped tasks.

The dataset contains 51,084 samples, 3 tasks, and has a reported 7.58% anomaly ratio in the test set.

Intended Use

This dataset is intended for research on:

• continual anomaly detection;
• continual learning for tabular data;
• predictive maintenance in industrial/automotive fleets;
• robustness under distribution shift;
• task ordering in continual-learning benchmarks;
• forgetting and knowledge transfer across related process-monitoring concepts;
• benchmarking anomaly detectors under sequential task exposure.

The intended use is defensive/industrial machine learning research, i.e. detecting component faults to support predictive maintenance.

Dataset Source

• SCANIA Component X Dataset: https://researchdata.se/en/catalogue/dataset/2024-34

The original dataset is a real-world (non-synthetic) predictive-maintenance dataset published by SCANIA CV AB and Stockholm University, describing an anonymized engine component's operational readouts, specifications, and repair labels across SCANIA trucks. Feature columns are anonymized identifiers (e.g. aa_000, ac_000, ...) per the source publisher's confidentiality requirements. Task boundaries in this benchmark are derived via spectral clustering over the original data, used as a proxy for concept drift.

Dataset Files

The repository contains the following files:

File	Description
data.csv	Main tabular dataset file.
orderings.json	Predefined task orderings for continual-learning evaluation.
croissant.json	Croissant metadata describing the dataset.

Dataset Structure

The main file is:

data.csv

The dataset contains task metadata, binary labels, and numerical, anonymized component-level operational/specification features.

Core Columns

Column	Type	Description
task_id	integer	Numeric identifier of the continual-learning task.
task_name	string	Name of the task, e.g. scania_0.
task_split	string	Split assignment for the row.
label	integer	Binary anomaly label. 0 denotes normal operation and 1 denotes a component anomaly/failure.

Task Identifiers

The dataset contains the following task identifiers:

scania_0, scania_1, scania_2

Feature Columns

The remaining 162 columns are anonymized numerical operational and specification variables from the original SCANIA Component X dataset, following the source publisher's anonymized naming scheme (two-letter code plus a numeric suffix). Representative examples include:

• aa_000, ac_000, ad_000, ae_000, af_000
• ag_000 … ag_009
• ay_000 … ay_009
• az_000 … az_009
• cn_000 … cn_009
• cs_000 … cs_009
• ee_000 … ee_009

For the complete schema, see croissant.json.

Task Orderings

The dataset provides six predefined orderings in orderings.json. These orderings define different continual-learning evaluation regimes over the same task set.

Ordering	Task sequence
curriculum_asc	scania_2 → scania_0 → scania_1
curriculum_desc	scania_1 → scania_0 → scania_2
generalization_desc	scania_1 → scania_2 → scania_0
generalization_asc	scania_0 → scania_2 → scania_1
smooth_drift	scania_0 → scania_1 → scania_2
abrupt_drift	scania_1 → scania_0 → scania_2

These orderings are intended to expose complementary continual-learning dynamics, including curriculum-like adaptation, generalization-oriented ordering, smooth drift, and abrupt drift.

Dataset Creation

The details of dataset creation can be found in our paper: link

Citation

When using the dataset, please cite:

@article{faber2026towards,
  title={Towards Principled Continual Anomaly Detection: A Systematic Framework and Benchmark Scenarios},
  author={Faber, Kamil and Smendowski, Mateusz and Corizzo, Roberto},
  journal={arXiv preprint arXiv:2607.18289},
  year={2026}
}

If you also use the underlying SCANIA Component X data, please additionally cite:

Kharazian, Z., Lindgren, T., Magnusson, S., Steinert, O., Andersson Reyna, O. (2025).
SCANIA Component X Dataset: A Real-World Multivariate Time Series Dataset for Predictive Maintenance.
Scientific Data. https://doi.org/10.1038/s41597-025-04802-6