nannyml.drift.model_inputs.univariate.statistical.calculator module

Statistical drift calculation using Kolmogorov-Smirnov and chi2-contingency tests.

class nannyml.drift.model_inputs.univariate.statistical.calculator.UnivariateStatisticalDriftCalculator(model_metadata: ModelMetadata, features: Optional[List[str]] = None, chunk_size: Optional[int] = None, chunk_number: Optional[int] = None, chunk_period: Optional[str] = None, chunker: Optional[Chunker] = None)[source]

Bases: DriftCalculator

A drift calculator that relies on statistics to detect drift.

Constructs a new UnivariateStatisticalDriftCalculator.

Parameters
  • model_metadata (ModelMetadata) – Metadata for the model whose data is to be processed.

  • features (List[str], default=None) – An optional list of feature names to use during drift calculation. None by default, in this case all features are used during calculation.

  • chunk_size (int) – Splits the data into chunks containing chunks_size observations. Only one of chunk_size, chunk_number or chunk_period should be given.

  • chunk_number (int) – Splits the data into chunk_number pieces. Only one of chunk_size, chunk_number or chunk_period should be given.

  • chunk_period (str) – Splits the data according to the given period. Only one of chunk_size, chunk_number or chunk_period should be given.

  • chunker (Chunker) – The Chunker used to split the data sets into a lists of chunks.

Examples

>>> import nannyml as nml
>>> ref_df, ana_df, _ = nml.load_synthetic_binary_classification_dataset()
>>> metadata = nml.extract_metadata(ref_df)
>>> # Create a calculator that will chunk by week
>>> drift_calc = nml.UnivariateStatisticalDriftCalculator(model_metadata=metadata, chunk_period='W')
calculate(data: DataFrame) UnivariateDriftResult[source]

Calculates the data reconstruction drift for a given data set.

Parameters

data (pd.DataFrame) – The dataset to calculate the reconstruction drift for.

Returns

reconstruction_drift – A result object where each row represents a Chunk, containing Chunk properties and the reconstruction_drift calculated for that Chunk.

Return type

UnivariateDriftResult

Examples

>>> import nannyml as nml
>>> ref_df, ana_df, _ = nml.load_synthetic_binary_classification_dataset()
>>> metadata = nml.extract_metadata(ref_df, model_type=nml.ModelType.CLASSIFICATION_BINARY)
>>> # Create a calculator and fit it
>>> drift_calc = nml.UnivariateStatisticalDriftCalculator(model_metadata=metadata, chunk_period='W').fit(ref_df)
>>> drift = drift_calc.calculate(data)
fit(reference_data: DataFrame)[source]

Fits the drift calculator using a set of reference data.

Parameters

reference_data (pd.DataFrame) – A reference data set containing predictions (labels and/or probabilities) and target values.

Returns

calculator – The fitted calculator.

Return type

DriftCalculator

Examples

>>> import nannyml as nml
>>> ref_df, ana_df, _ = nml.load_synthetic_binary_classification_dataset()
>>> metadata = nml.extract_metadata(ref_df, model_type=nml.ModelType.CLASSIFICATION_BINARY)
>>> # Create a calculator and fit it
>>> drift_calc = nml.UnivariateStatisticalDriftCalculator(model_metadata=metadata, chunk_period='W').fit(ref_df)