load_ecg_diff_count_3

load_ecg_diff_count_3(learning_type: Literal['unsupervised', 'semi-supervised', 'supervised'] = 'unsupervised') → tuple[ndarray, ndarray] | tuple[ndarray, ndarray, ndarray, ndarray]

Load the synthetic ECG dataset ‘ecg-diff-count-3’.

The dataset contains three different kind of anomalies. The dataset was generated using GutenTAG [1]

Parameters:

learning_typestr, default = “unsupervised”

The learning type of the dataset. Must be one of “unsupervised”, “semi-supervised”, or “supervised”. If “unsupervised”, only the test partition is loaded. If “semi-supervised”, the test partition and the training partition without anomalies is returned. If “supervised”, the training partition with anomalies is returned instead.

Returns:

X_testnp.ndarray

Multivariate test time series with shape (10000,2).

y_testnp.ndarray

Binary anomaly labels for the test time series with shape (10000,).

X_trainnp.ndarray, optional

Multivariate train time series with shape (10000,2). Omitted if learning_type is “unsupervised”.

y_trainnp.ndarray, optional

Binary anomaly labels for the train time series with shape (10000,). Omitted if learning_type is “unsupervised”.

Notes

Dimensionality: univariate Series length: 10000 Frequency: unknown Learning Type: all supported

References

[1]

Phillip Wenig, Sebastian Schmidl, and Thorsten Papenbrock. TimeEval: A Benchmarking Toolkit for Time Series Anomaly Detection Algorithms. PVLDB, 15(12): 3678 - 3681, 2022. doi:10.14778/3554821.3554873.

Examples

>>> from aeon.datasets import load_ecg_diff_count_3
>>> X_test, y_test, X_train, y_train = load_ecg_diff_count_3("supervised")