Multivariate time series anomaly detection for shift quality calibration.
Please set the parameters on config.ini
before you do anything.
If you run the data_split.py
, you can get the train, validation, and test set for anomaly detection.
If the data length is shorter than time window, it will drop.