cascade.utils.sklearn#

class cascade.utils.sklearn.SkMetric(name: str, *args: Any, value: Optional[SupportsFloat] = None, dataset: Optional[str] = None, split: Optional[str] = None, direction: Optional[Literal['up', 'down']] = None, interval: Optional[Tuple[SupportsFloat, SupportsFloat]] = None, extra: Optional[Dict[str, SupportsFloat]] = None, **kwargs: Any)[source]#

__init__(name: str, *args: Any, value: Optional[SupportsFloat] = None, dataset: Optional[str] = None, split: Optional[str] = None, direction: Optional[Literal['up', 'down']] = None, interval: Optional[Tuple[SupportsFloat, SupportsFloat]] = None, extra: Optional[Dict[str, SupportsFloat]] = None, **kwargs: Any) → None[source]#

Cascade Metric wrapper for sklearn.metrics

Parameters:

name (str) – name of the metric, should be an exact match of the members of sklearn.metrics, but you can also use the aliases from the example
value (Optional[MetricType], optional) – Can assign value to the metric if already computed, by default None
dataset (Optional[str], optional) – Name of the dataset metric was computed on, by default None
split (Optional[str], optional) – Name of the dataset split like train, test or eval, by default None
direction (Optional[Literal["up", "down"]], optional) – Select up if greater the better or down if worse the better, by default None
interval (Optional[Tuple[MetricType, MetricType]], optional) – Confidence interval, by default None
extra (Optional[Dict[str, MetricType]], optional) – Extra metadata for a metric, by default None

Example

Here the aliases are demonstrated. You can also use any names from sklearn.metrics

                          from cascade.utils.sklearn import SkMetric

acc = SkMetric("acc")
precision = SkMetric("precision")
recall = SkMetric("recall")
mse = SkMetric("mse")

gt_pred = ([0, 0], [0, 1])

acc.compute(*gt_pred)
precision.compute(*gt_pred)
recall.compute(*gt_pred)
mse.compute(*gt_pred)

                        

                          >>> from pprint import pprint
>>> from cascade.utils.sklearn.sk_metric import METRIC_ALIASES
>>> pprint(METRIC_ALIASES)
{'acc': 'accuracy_score',
 'accuracy': 'accuracy_score',
 'f1': 'f1_score',
 'mae': 'mean_absolute_error',
 'mse': 'mean_squared_error',
 'precision': 'precision_score',
 'recall': 'recall_score'}

                        

compute(*args: Any, **kwargs: Any) → SupportsFloat[source]#: The method to compute metric’s value. Should always populate the internal self.value field and return it.

class cascade.utils.sklearn.SkModel(*args: Any, blocks: Optional[List[Any]] = None, **kwargs: Any)[source]#

Wrapper for sklearn models. Accepts the name and block to form pipeline. Can fit, evaluate, predict save and load out of the box.

__init__(*args: Any, blocks: Optional[List[Any]] = None, **kwargs: Any) → None[source]#

Parameters:: blocks (list, optional) – List of sklearn transformers to make a pipeline from

fit(*args: Any, **kwargs: Any) → None[source]#: Wrapper for pipeline.fit

get_meta() → List[Dict[Any, Any]][source]#

Returns:

meta – A list where first element is this object’s metadata. All other elements represent the other stages of pipeline if present.

Meta can be anything that is worth to document about the object and its properties.

Meta is a list (see Meta type alias) to allow the formation of pipelines.

Return type: