Features.retain

Features.retain(toRetain=None, start=None, end=None, number=None, randomize=False, points=None, *, useLog=None)

Keep only certain features of this object.

A variety of methods for specifying features to keep based on the provided parameters. If toRetain is not None, start and end must be None. If start or end is not None, toRetain must be None.

The nimble.match module contains many helpful functions that could be used for toRetain.

Parameters:

  • toRetain (identifier, list of identifiers, function, query) –

    • identifier - a name or index

    • list of identifiers - an iterable container of identifiers

    • function - accepts a feature as its only argument and returns a boolean value to indicate if the feature should be retained. See nimble.match for common functions.

    • query - string in the format ‘POINTNAME OPERATOR VALUE’ (i.e “pt1 < 10”, “id4 == yes”, or “row4 is nonZero”) where OPERATOR is separated from the POINTNAME and VALUE by whitespace characters. See nimble.match.QueryString for string requirements.

  • start (identifier) – Parameters indicating range based retention. Begin the retention at the location of start. Finish retaining at the inclusive end location. If only one of start and end are non-None, the other default to 0 and the number of values in each feature, respectively.

  • end (identifier) – Parameters indicating range based retention. Begin the retention at the location of start. Finish retaining at the inclusive end location. If only one of start and end are non-None, the other default to 0 and the number of values in each feature, respectively.

  • number (int) – The quantity of features that are to be retained, the default None means unrestricted retained. This can be provided on its own (toRetain, start and end are None) to the first number of features, or in conjuction with toRetain or start and end, to limit their output.

  • randomize (bool) – Indicates whether random sampling is to be used in conjunction with the number parameter. If randomize is False, the chosen features are determined by feature order, otherwise it is uniform random across the space of possible features.

  • points (None, identifier, list of identifiers) – Only apply the target function to a selection of points in each feature. May be a single point name or index, an iterable, container of point names and/or indices. None indicates application to all points.

  • useLog (bool, None) – Local control for whether to send object creation to the logger. If None (default), use the value as specified in the “logger” “enabledByDefault” configuration option. If True, send to the logger regardless of the global option. If False, do NOT send to the logger, regardless of the global option.

See also

extract, delete

Examples

Retain a single feature.

>>> X = nimble.identity(3)
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain('a')
>>> X
<Matrix 3pt x 1ft
     a
   ┌──
 0 │ 1
 1 │ 0
 2 │ 0
>

Retain multiple features.

>>> X = nimble.identity(3)
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain(['a', 2])
>>> X
<Matrix 3pt x 2ft
     a  c
   ┌─────
 0 │ 1  0
 1 │ 0  0
 2 │ 0  1
>

Retain feature when the function returns True.

>>> X = nimble.data([[1, None, 3], [None, 5, 6], [7, 8, 9]])
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain(nimble.match.allNonMissing)
>>> X
<Matrix 3pt x 1ft
       c
   ┌──────
 0 │ 3.000
 1 │ 6.000
 2 │ 9.000
>

Retain feature when the query string returns True.

>>> X = nimble.identity(3, featureNames=['a', 'b', 'c'],
...                     pointNames=['p1', 'p2', 'p3'])
>>> X.features.retain("p2 != 0")
>>> X
<Matrix 3pt x 1ft
      b
    ┌──
 p1 │ 0
 p2 │ 1
 p3 │ 0
>

Retain features from the inclusive start to the inclusive end.

>>> X = nimble.identity(3)
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain(start=1, end=2)
>>> X
<Matrix 3pt x 2ft
     b  c
   ┌─────
 0 │ 0  0
 1 │ 1  0
 2 │ 0  1
>

Select a set number to retain, starting from the first feature.

>>> X = nimble.identity(3)
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain(number=2)
>>> X
<Matrix 3pt x 2ft
     a  b
   ┌─────
 0 │ 1  0
 1 │ 0  1
 2 │ 0  0
>

Select a set number to retain, choosing features at random.

>>> nimble.random.setSeed(42)
>>> X = nimble.identity(3)
>>> X.features.setNames(['a', 'b', 'c'])
>>> X.features.retain(number=2, randomize=True)
>>> X
<Matrix 3pt x 2ft
     c  a
   ┌─────
 0 │ 0  1
 1 │ 0  0
 2 │ 1  0
>

Keywords: keep, hold, maintain, preserve, remove, filter