Skip to content

rockfish.actions.replace

Attributes

Condition = Union[EqualizeCondition, TopKCondition, ThresholdCondition, SQLCondition] module-attribute

Resample = Union[ValuesResample, SQLResample] module-attribute

Classes

ReplaceConfig

Configuration class for the Replace action.

Attributes:

Name Type Description
field str

Field name for replacement.
condition Condition

The condition that defines the criteria for selecting values to be replaced. Options include: - EqualizeCondition - TopKCondition - ThresholdCondition - SQLCondition
resample Optional[Resample]

The resample strategy used to obtain values and weights for replacement. If not provided, the retained values are resampled using their frequencies as weights. Options include: - ValuesResample - SQLResample
seed Optional[int]

Seed for random number generator.

Replace

Replace values in a selected field with given condition with new values from the resampled result.

Attributes:

Name Type Description
Config

Alias for ReplaceConfig.

EqualizeCondition

EqualizeCondition class for the condition configuration.

Attributes:

Name Type Description
equalization bool

If equalization is True, it enables equalization, replacing frequent values with rare ones to achieve equal rates. The value of resample in ReplaceConfig will be ignored if equalization is enabled.

TopKCondition

TopKCondition class for the condition configuration.

Attributes:

Name Type Description
top_k int

The number of top values to retain. The values not in the top_k will be replaced.

ThresholdCondition

ThresholdCondition class for the condition configuration.

Attributes:

Name Type Description
threshold float

The threshold value to retain. The values occupying less than the threshold will be replaced.

SQLCondition

SQLCondition class for the condition configuration.

Attributes:

Name Type Description
query str

A SQL query string that returns a 'mask' column. The values with True in the 'mask' column will be replaced. The input dataset is always referred to as my_table in the query.

ValuesResample

ValuesResample class for the resample configuration.

Attributes:

Name Type Description
replace_values list

A list of values to replace. The type of these values must match the field type. Each value is automatically assigned a weight of 1.

SQLResample

SQLResample class for the resample configuration.

Attributes:

Name Type Description
query str

A SQL query string that returns 'values' and 'weights' columns. The 'values' column contains the values to replace, and the 'weights' column contains the weights for each value. The input dataset is always referred to as my_table in the query.