deephyper.analysis.hpo.parameters_at_max

deephyper.analysis.hpo.parameters_at_max#

deephyper.analysis.hpo.parameters_at_max(df: DataFrame, column: str = 'objective', prefix: str = 'p:', n_last: int | float = 1.0) Tuple[dict, float][source]#

Return the parameters at the maximum of the given column function.

Parameters:
  • df (pd.DataFrame) – the results of a Hyperparameter Search.

  • column (str) – the column to use for the maximization. Defaults to "objective".

  • prefix (str) – the prefix of columns of parameters.

  • n_last (int | float) – Select the max among the n_last rows. If it is a float it is among the int(n_last * len(df)) last rows.

Returns:

the parameters at the maximum of the column and its corresponding

value.

Return type:

Tuple[dict, float]