paddlets.utils.utils

check_model_fitted(model: Trainable, msg: Optional[str] = None)[source]

check if model has fitted, Raise Exception if not fitted

Parameters

  • model (Trainable) – model instance.

  • msg (str) – str, default=None The default error message is, “This %(name)s instance is not fitted yet. Call ‘fit’ with appropriate arguments before using this estimator.” For custom messages if “%(name)s” is present in the message string, it is substituted for the estimator name. Eg. : “Estimator, %(name)s, must be fitted before sparsifying”.

Returns

None

Raises

ValueError –

get_uuid(prefix: str = '', suffix: str = '')[source]

Get a random string of 16 characters.

Parameters

  • prefix (str, optional) – The prefix of the returned string.

  • suffix (str, optional) – The suffix of the returned string.

Returns

String of 16 characters.

Return type

str

month_delta(start_date, end_date)[source]

Month delta

Parameters

  • start_date (datetime) – start date

  • end_date (datetime) – end date

week_delta(start_date, end_date)[source]

Week_delta

Parameters

  • start_date (datetime) – start date

  • end_date (datetime) – end date

check_train_valid_continuity(train_data: TSDataset, valid_data: TSDataset) → bool[source]

Check if train and test TSDataset are continous

Parameters
Returns

if train and test TSDataset are continous

Return type

bool

split_dataset(dataset: TSDataset, split_point: int) → TSDataset[source]

Split dataset (accroding to the max length)

Parameters

  • dataset (TSDataset) – dataset to be splited.

  • split_point (int) – split point.

Returns

TSDataset

get_tsdataset_max_len(dataset: TSDataset) → int[source]

Get dataset max length

Parameters

dataset (TSDataset) – dataset use to get length.

Returns

int

repr_results_to_tsdataset(reprs: array, dataset: TSDataset) → TSDataset[source]

Convert representation model output to a TSDataset

Parameters

  • reprs (np.array) – output results of representation model

  • dataset (TSDataset) – dataset use to get target

Returns

TSDataset

plot_anoms(predict_data: Optional[TSDataset] = None, origin_data: Optional[TSDataset] = None, feature_name: Optional[str] = None)[source]

Plots anomalies

Parameters

  • predict_data (TSDataset) – Data used to print predict anom labels.

  • origin_data (TSDataset|None) – Data used to print features or origin anom labels. only print predict anom labels if set to None.

  • feature_name (str|None) – feature name in origin data to print

build_ts_infer_input(tsdataset: TSDataset, meta_file: str) → Dict[str, ndarray][source]

Build time series input for infer tensor base on TSDataset and meta_file which is generated by paddlets.model.save(…, network_model=True, …).

Parameters

  • tsdataset (TSDataset) – The time series dataset.

  • meta_file (str) – The Meta file which is generated by paddlets.model.save(…, network_model=True, …).

Returns

The np.ndarray dict which is match with paddle_infer.input_names.

Return type

Dict[str, np.ndarray]