Transformer

ColumnSelect

Column selection transformer.

In transform, subsets X to columns provided as hyper-parameters.

Sequence of columns in Xt=transform(X) is as in columns hyper-parameter. Caveat: this means that transform may change sequence of columns,

even if no columns are removed from X in transform(X).

Schnellstart

python

from sktime.transformations.subset import ColumnSelect

estimator = ColumnSelect(columns=None, integer_treatment='col', index_treatment='remove')

Parameter(3)

columnspandas compatible index or index coercible, optional, default = None: columns to which X in transform is to be subset
integer_treatmentstr, optional, one of “col” (default) and “coerce”: determines how integer index columns are treated “col” = subsets by column iloc index, even if columns is not in X.columns “coerce” = coerces to integer pandas.Index and attempts to subset
index_treatmentstr, optional, one of “remove” (default) or “keep”: determines which column are kept in Xt = transform(X, y) “remove” = only indices that appear in both X and columns are present in Xt. “keep” = all indices in columns appear in Xt. If not present in X, NA is filled.

Beispiele

>>> from sktime.transformations.subset import ColumnSelect
>>> from sktime.datasets import load_longley
>>> X = load_longley ()[1 ]
>>> transformer = ColumnSelect (columns = ["GNPDEFL", "POP", "FOO" ])
>>> X_subset = transformer. fit_transform (X = X)