{
    "name": "CSTSNonnormalTrain",
    "source": "https://huggingface.co/datasets/idegen/csts",
    "desp": "CSTS (Correlation Structures in Time Series) is a comprehensive synthetic benchmarking dataset for evaluating correlation structure discovery in time series data. This dataset is a subset of the full CSTS training dataset, with distribution shift.",
    "domain": "Others",
    "license": "CC BY 4.0",
    "targets": [
        "iob",
        "cob",
        "ig"
    ],
    "covariates": [],
    "timestamp": "timestamp",
    "freq": [
        "s"
    ],
    "num_series": 30,
    "num_timesteps": 37920300,
    "num_datapoints": 151681200,
    "split": "train",
    "quality": "very high",
    "timestamp_as_covariate": true
}