deephyper.analysis.hpo.rank#

deephyper.analysis.hpo.rank(a, method='min', decimals=3, *, axis=None, nan_policy='propagate')[source]#

Returns the ranking from a list of scores given a tolerance epsilon (wrapper around scipy.stats.rankdata, see Scipy Documentation). Lower scores corresponds to lower ranks.

Parameters:
  • a (array) – List of scores.

  • method (str, optional) – The method used to assign ranks to tied elements. The options are "average", "min", "max", "dense" and 'ordinal'. Defaults to "min".

  • decimals (int, optional) – The number of decimal at which rounding is performed. Defaults to 3.

  • axis (int, optional) – The axis along which the elements of a are ranked. Defaults to None to rank the elements after flattening the array.

  • nan_policy (str, optional) – Defines how to handle when input contains nan. The options are "propagate", "raise", "omit". Defaults to "propagate".

Returns:

The ranking of the scores.

Return type:

array