The results of an attribution explanation run by the Fiddler engine
| Attribute | Type | Description |
|---|---|---|
| algorithm | str | The name of the explanation method |
| inputs | List[str] | List of input features |
| attributions | List[float] | List of attributions |
| misc | Optional[dict] | misc will have different details based on the explanation type. For example model prediction, baseline prediction, confidence interval for the explanation, size of background dataset, ... |
AttributionExplanation(
algorithm='fiddler_shapley_values',
inputs=['Age', 'Balance', 'CreditScore', 'EstimatedSalary',
'Gender', 'Geography', 'HasCrCard', 'IsActiveMember',
'NumOfProducts', 'Tenure'],
attributions=[0.15160113491423066, 0.031480978762930156,
0.1831941120167941, -0.0070094998168561155,
-0.003002888334429402, 0.029338303601551853,
-0.005787066742637167, -0.0452038935301122,
0.14073604489745767, -0.012805550127150742],
misc={'background_dataset_size': 200,
'baseline_prediction': 0.18426200542835106,
'explanation_ci': {'Age': 0.012645587534114936,
'Balance': 0.009587602106958078,
'CreditScore': 0.009863703930511359,
'EstimatedSalary': 0.0023364778612537935,
'Gender': 0.0016714116997844039,
'Geography': 0.006029320586649526,
'HasCrCard': 0.002847070811544292,
'IsActiveMember': 0.008268189814243227,
'NumOfProducts': 0.02431611472372665,
'Tenure': 0.0022312720267721868},
'explanation_ci_level': 0.95,
'model_prediction': 0.6468036810701299
}
)