Published July 3, 2025 | Version v1
Dataset Open

Synthetic Anomaly Test Sets from CASAS ARUBA and HHL120 Datasets

Authors/Creators

  • 1. ROR icon University of Castilla-La Mancha

Description

This dataset has been designed to support research on detecting daily behavioural anomalies in older adults using non-invasive ambient sensor data. It contains test subsets derived from the publicly available CASAS ARUBA and HHL120 smart home datasets, with synthetic anomalies injected to simulate unusual activities or patterns.

The data is organised into two main folders:

  • Aruba

  • HHL120

Each folder contains subfolders corresponding to different areas of the home: Bathroom, Bedroom, Kitchen, Living Room

For each area, you will find:

  • test_anomalia_{area}: Test datasets with synthetic anomalies injected.

  • y_true_{area}: Ground-truth labels indicating the specific days when anomalies were introduced.

These test sets are intended to serve as reproducible benchmarks for evaluating and comparing anomaly detection models in the smart home domain.

The original training data can be freely accessed at: https://data.casas.wsu.edu/

Please cite the original CASAS project when using this dataset in your research.

Files

CASAS.zip

Files (89.5 kB)

Name Size Download all
md5:2cb8122cb47f565be172119a8d14942f
89.5 kB Preview Download

Additional details

Funding

Ministerio de Ciencia, Innovación y Universidades
TED2021-132149B-C41