Timeseries Explainability Bug - 2.214.0

[Alex-Wenner-FHR]

Describe the bug

• From the documentation and the Schemas, I think I am passing in all of the right information, with the correct types. For whatever reason I am hitting an error giving a schema validation on two keys that are indeed called out in the docs.

analyzer.utils.exceptions.CustomerError: Analysis configuration schema validation error: Key 'time_series_data_config' error: Wrong keys 'dataset_format', 'static_covariates

during the explainability job -

clarify_processor.run_explainability(
    data_config = data_config,
    model_config = model_config,
    explainability_config = asym_shap_val_config,
)

Here is an example of my TimeSeriesDataConfig:

ts_data_config = TimeSeriesDataConfig(
    target_time_series = "[*].target",
    item_id = "[*].source",
    timestamp = "[*].timestamp",
    related_time_series = [f"[*].{r}" for r in related_timeseries],
    static_covariates= [f"[*].{r}" for r in dataset_generation_spec["static_feature_columns"]],
    dataset_format= TimeSeriesJSONDatasetFormat.TIMESTAMP_RECORDS
)
ts_data_config.get_time_series_data_config()

This .get_time_series_data_config() outputs something as follows:

{'target_time_series': '[*].target',
 'item_id': '[*].source',
 'timestamp': '[*].timestamp',
 'related_time_series': ['[*].rt1',
  '[*].rt2',
  '[*].rt3',
  '[*].rt4',
  '[*].rt5',
  '[*].rt6',
  '[*].rt7',
  '[*].rt8',
  '[*].rt9',
  '[*].rt10',
  '[*].rt11',
  '[*].rt12',
  '[*].rt13',
  '[*].rt15',
  '[*].rt16',
  '[*].rt17'],
 'static_covariates': ['[*].sc1',
  '[*].sc2',
  '[*].sc3',
  '[*].sc4'],
 'dataset_format': 'timestamp_records'}

System information

• SageMaker Python SDK version: 2.214.0
• Framework name (eg. PyTorch) or algorithm (eg. KMeans): N/A
• Framework version: N/A
• Python version: 3.11.0
• CPU or GPU: CPU
• Custom Docker image (Y/N): N/A

[rvasahu-amazon]

Hi Alex, thanks for reaching out. Although we've merged the SDK changes to support time series explainability, the actual feature is not released at the moment. We'll update you when it's available.

[Alex-Wenner-FHR]

Please do @rvasahu-amazon this is a high demand feature! Thanks.