freneticlib.core.objective

Module Contents

Classes

Objective

Objective base class. Don't use.

MaxObjective

For the specification to maximize a given feature.

MinObjective

For the specification to minimize a given feature.
class freneticlib.core.objective.Objective(feature, per_simulation_aggregator: Callable | str | List | Dict, threshold: float = None, dynamic_threshold_quantile: float = None)[source]

Bases: abc.ABC

Objective base class. Don’t use.

Parameters:

  • feature (str) – The feature that should be optimised.

  • per_simulation_aggregator (Union[Callable, str, List, Dict]) – How to aggregate over a simulation’s records. (see https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.aggregate.html)

  • threshold (float, optional) – Only consider elements “better” than this. Defaults to None

  • dynamic_threshold_quantile (float, optional) – If set, specifies which quantile of values should be used for re-calculation of threshold.

get_best(df: pandas.DataFrame) → pandas.Series[source]

Returns the row whose <feature> value is best (i.e. maximal or minimal). :param df: The execution history. :type df: pd.DataFrame

Returns:

The best row.

Return type:

(pd.Series)

abstract recalculate_dynamic_threshold(df: pandas.DataFrame)[source]

Recalculates the dynamic threshold, if it is provided.

Parameters:

df (pd.DataFrame) – The execution history.

abstract filter_by_threshold(df: pandas.DataFrame) → pandas.DataFrame[source]

If threshold is specified, filters the dataframe to only contain rows where the feature value exceeds the threshold.

Parameters:

df (pd.DataFrame) – The execution history.

Returns:

The filtered execution history.

Return type:

df (pd.DataFrame)

class freneticlib.core.objective.MaxObjective(*args, **kwargs)[source]

Bases: Objective

For the specification to maximize a given feature.

See AbstractObject.__init__ for parameters.

filter_by_threshold(df: pandas.DataFrame) → pandas.DataFrame[source]

If threshold is specified, filters the dataframe to only contain rows where self.feature >= self.threshold.

Parameters:

df (pd.DataFrame) – The execution history.

Returns:

The filtered execution history.

Return type:

df (pd.DataFrame)

recalculate_dynamic_threshold(df: pandas.DataFrame)[source]

Recalculates the dynamic threshold according to self.dynamic_threshold_quantile and updates it if the new value is higher than the previous threshold.

Parameters:

df (pd.DataFrame) – The execution history.

class freneticlib.core.objective.MinObjective(*args, **kwargs)[source]

Bases: Objective

For the specification to minimize a given feature.

Constructor for MinObjective. Forwards arguments to Objective.

Parameters:

  • *args – Arguments that are forwarded to super constructor.

  • **kwargs – KW-arguments that are forwarded to super constructor.

filter_by_threshold(df: pandas.DataFrame) → pandas.DataFrame[source]

If threshold is specified, filters the dataframe to only contain rows where self.feature <= self.threshold.

Parameters:

df (pd.DataFrame) – The execution history.

Returns:

The filtered execution history.

Return type:

df (pd.DataFrame)

recalculate_dynamic_threshold(df: pandas.DataFrame)[source]

Recalculates the dynamic threshold according to self.dynamic_threshold_quantile and updates it if the new value is lower than the previous threshold.

Parameters:

df (pd.DataFrame) – The execution history.