RESCALE_OUTLIERS_WITH_Z_SCORE
Returns a new column with a rescaled outlier value in each row, based on the settings in the parameters. This action also applies Z-score normalization to linearly scale data values to have a mean (μ) of 0 and standard deviation (σ) of 1. We recommend this action for handling outliers.
Parameters
-
sourceColumn
– Specifies the name of an existing numeric column that might contain outliers. -
targetColumn
– Specifies the name of an existing numeric column that might contain outliers. -
outlierStrategy
– Specifies the approach to use in detecting outliers. Valid values include the following:-
Z_SCORE
– Identifies a value as an outlier when it deviates from the mean by more than the standard deviation threshold. -
MODIFIED_Z_SCORE
– Identifies a value as an outlier when it deviates from the median by more than the median absolute deviation threshold. -
IQR
– Identifies a values as an outlier when it falls beyond the first and last quartile of column data. The interquartile range (IQR) measures where the middle 50% of the data points are.
-
-
threshold
– The threshold value to use when detecting outliers. ThesourceColumn
value is identified as an outlier if the score that's calculated with theoutlierStrategy
exceeds this number. The default is 3.
The following examples display syntax for a single RecipeAction operation. A recipe contains at least one RecipeStep operation, and a recipe step contains at least one recipe action. A recipe action runs the data transform that you specify. A group of recipe actions run in sequential order to create the final dataset.