scipy.stats.rankdata#

scipy.stats.rankdata(a, method='average', *, axis=None, nan_policy='propagate')[source]#

Assign ranks to data, dealing with ties appropriately.

By default (axis=None), the data array is first flattened, and a flat array of ranks is returned. Separately reshape the rank array to the shape of the data array if desired (see Examples).

Ranks begin at 1. The method argument controls how ranks are assigned to equal values. See [1] for further discussion of ranking methods.

Parameters:
aarray_like

The array of values to be ranked.

method{‘average’, ‘min’, ‘max’, ‘dense’, ‘ordinal’}, optional

The method used to assign ranks to tied elements. The following methods are available (default is ‘average’):

  • ‘average’: The average of the ranks that would have been assigned to all the tied values is assigned to each value.

  • ‘min’: The minimum of the ranks that would have been assigned to all the tied values is assigned to each value. (This is also referred to as “competition” ranking.)

  • ‘max’: The maximum of the ranks that would have been assigned to all the tied values is assigned to each value.

  • ‘dense’: Like ‘min’, but the rank of the next highest element is assigned the rank immediately after those assigned to the tied elements.

  • ‘ordinal’: All values are given a distinct rank, corresponding to the order that the values occur in a.

axis{None, int}, optional

Axis along which to perform the ranking. If None, the data array is first flattened.

nan_policy{‘propagate’, ‘omit’, ‘raise’}, optional

Defines how to handle when input contains nan. The following options are available (default is ‘propagate’):

  • ‘propagate’: propagates nans through the rank calculation

  • ‘omit’: performs the calculations ignoring nan values

  • ‘raise’: raises an error

Note

When nan_policy is ‘propagate’, the output is an array of all nans because ranks relative to nans in the input are undefined. When nan_policy is ‘omit’, nans in a are ignored when ranking the other values, and the corresponding locations of the output are nan. This behavior is the default because it is intuitive and compatible with the behavior before the nan_policy parameter was introduced.

New in version 1.10.

Returns:
ranksndarray

An array of size equal to the size of a, containing rank scores.

References

Examples

>>> import numpy as np
>>> from scipy.stats import rankdata
>>> rankdata([0, 2, 3, 2])
array([ 1. ,  2.5,  4. ,  2.5])
>>> rankdata([0, 2, 3, 2], method='min')
array([ 1,  2,  4,  2])
>>> rankdata([0, 2, 3, 2], method='max')
array([ 1,  3,  4,  3])
>>> rankdata([0, 2, 3, 2], method='dense')
array([ 1,  2,  3,  2])
>>> rankdata([0, 2, 3, 2], method='ordinal')
array([ 1,  2,  4,  3])
>>> rankdata([[0, 2], [3, 2]]).reshape(2,2)
array([[1. , 2.5],
      [4. , 2.5]])
>>> rankdata([[0, 2, 2], [3, 2, 5]], axis=1)
array([[1. , 2.5, 2.5],
       [2. , 1. , 3. ]])
>>> rankdata([0, 2, 3, np.nan, -2, np.nan], nan_policy="propagate")
array([nan, nan, nan, nan, nan, nan])
>>> rankdata([0, 2, 3, np.nan, -2, np.nan], nan_policy="omit")
array([ 2.,  3.,  4., nan,  1., nan])