sklearnFeatureTransformer¶

This class is a subclass of BaseFeatureTransformer and is used for transforming input data using scikit-learn transformers. It provides methods for fitting and transforming the data.

Configuration¶

Required Configuration¶

sklearnFeatureTransformer contains the following required configuration:

transformers: A list of transformers to run. Each transformer is a dictionary with the following entries:
- transformer: The sklearn.preprocessing class to use as the transformer. Required.
- transformer_args: A list of arguments to pass into the transformer. Optional.
- transformer_kwargs: A dictionary of keyword arguments to pass into the transformer. Optional.
- transformer_columns: A list of columns names to apply the transformer to.

Optional Configuration¶

sklearnFeatureTransformer contains no optional configuration.

Default Configuration¶

sklearnFeatureTransformer contains the following default configuration:

column_transformer_kwargs: A dictionary of keyword arguments to pass into the ColumnTransformer class. By default, this contains {'remainder': 'passthrough'}, which explicitly instructs the ColumnTransformer to pass through any columns in the dataset untransformed that are not mentioned in the transformers dictionary. If you override this, note that the default behavior for ColumnTransformer is to drop these columns.

Attributes¶

transformer : An instance of sklearn.compose.ColumnTransformer, which is used to run all the transformers provided in the configuraiton.

Methods¶

fit¶

This method fits the feature transformer to the input data.

def fit(self, data, *args, **kwargs)

Arguments:

data (array-like): The input data to fit the transformer to.
**kwargs (optional): Keyword arguments to be passed to the fit method.

Returns:

self.transformer: The fitted ColumnTransformer object.

transform¶

This method applies the fitted feature transformer to transform the input data.

def transform(self, data, *args, **kwargs)

Arguments:

data (array-like): The input data to be transformed.
**kwargs (optional): Keyword arguments to be passed to the transform method.

Returns:

numpy.ndarray or scipy.sparse matrix: The transformed data.

fit_transform¶

This method fits the feature transformer to the data and transforms it.

def fit_transform(self, data, *args, **kwargs)

Arguments:

data (array-like): The input data to fit the transformer to and transform.
**kwargs (optional): Keyword arguments to be passed to the fit_transform method.

Returns:

numpy.ndarray or scipy.sparse matrix: The transformed data.

_get_transformer¶

This method gets the scikit-learn transformer object based on the given class name.

def _get_transformer(self, transformer_class, *args, **kwargs)

Arguments:

transformer_class: The class name of the transformer.
**kwargs (optional): Keyword arguments to be passed to the transformer constructor.

Returns:

object: The instantiated scikit-learn transformer object.

Usage:¶

from lolpop.component import sklearnFeatureTransformer 

data = #generate data 

conf = {
    #insert component configuration here ... 

    "config" : {
        "transformers" : [
            {"transformer": "MinMaxScaler",
            "transformer_columns": ["height", "weight"]},
            {"transformer": "OneHotEncoder",
            "transformer_columns": ["sex", "state"]},
        ]
    }
}

transformer = sklearnFeatureTransformer(conf=conf)
transformed_data = transformer.fit_transform(data)

print(transformed_data)

Sample Configuration¶

conf = {
    "config": {
        "transformers": 
        [
        {"transformer": "OneHotEncoder", "transformer_columns": ["Sex"]},
        {"transformer": "MinMaxScaler", "transformer_columns": ["Height", "Weight", "Diameter"]},
        ]
    }
}