dataeval.quality.PrioritizeOutput¶

class dataeval.quality.PrioritizeOutput¶

Output class for Prioritize quality evaluator.

indices¶

Indices that sort the dataset in order of priority according to the specified method and policy. These indices can be used with the Indices selection class.

Type:¶

NDArray[np.intp]

scores¶

Prioritization scores for each sample (only available for methods that compute scores: “knn” and “kmeans_distance”). Scores are ordered according to the original dataset order, not the prioritized order.

Type:¶

NDArray[np.float32] | None

method¶

The prioritization method that was used.

Type:¶

Literal[“knn”, “kmeans_distance”, “kmeans_complexity”]

policy¶

The selection policy that was applied.

Type:¶

Literal[“hard_first”, “easy_first”, “stratified”, “class_balance”]

data()¶

Returns the prioritized indices.

Return type:¶

numpy.typing.NDArray[numpy.intp]