facet.simulation.UnivariateSimulationResult#

class facet.simulation.UnivariateSimulationResult(*, partitioner, mean, sem, feature_name, output_name, output_unit, baseline, confidence_level)[source]#

Summary result of a univariate simulation.

Generic types

~T_Values(bound= generic)

Parameters

partitioner (Partitioner[TypeVar(T_Values, bound= generic)]) – the partitioner used to generate feature values to be simulated
mean (Sequence[float]) – mean predictions for the values representing each partition
sem (Sequence[float]) – standard errors of the mean predictions for the values representing each partition
feature_name (str) – name of the simulated feature
output_name (str) – name of the target for which outputs are simulated
output_unit (str) – the unit of the simulated outputs (e.g., uplift or class probability)
baseline (float) – the average observed actual output, acting as the baseline of the simulation
confidence_level (float) – the width of the confidence interval determined by the standard error of the mean, ranging between 0.0 and 1.0 (exclusive)

Attribute summary

`COL_LOWER_BOUND`	The name of a series of lower CI bounds of simulated values per partition.
`COL_MEAN`	The name of a series of mean simulated values per partition.
`COL_SEM`	The name of a series of standard errors of mean simulated values per partition.
`COL_UPPER_BOUND`	The name of a series of upper CI bounds of simulated values per partition.
`IDX_PARTITION`	The name of the column index of attribute `output`, denoting partitions represented by their central values or by a category.
`data`	The simulation result as a data frame, indexed by the central values of the partitions for which the simulation was run, with the following columns:
`partitioner`	The partitioner used to generate feature values to be simulated.
`feature_name`	Name of the simulated feature.
`output_name`	Name of the target for which outputs are simulated.
`output_unit`	The unit of the simulated outputs (e.g., uplift or class probability).
`baseline`	The average observed actual output, acting as the baseline of the simulation.
`confidence_level`	The width \(\alpha\) of the confidence interval determined by bootstrapping, with \(0 < \alpha < 1\).

Definitions

COL_LOWER_BOUND = 'lower_bound'#: The name of a series of lower CI bounds of simulated values per partition.

COL_MEAN = 'mean'#: The name of a series of mean simulated values per partition.

COL_SEM = 'sem'#: The name of a series of standard errors of mean simulated values per partition.

COL_UPPER_BOUND = 'upper_bound'#: The name of a series of upper CI bounds of simulated values per partition.

IDX_PARTITION = 'partition'#: The name of the column index of attribute output, denoting partitions represented by their central values or by a category.

baseline: float#: The average observed actual output, acting as the baseline of the simulation.

confidence_level: float#: The width \(\alpha\) of the confidence interval determined by bootstrapping, with \(0 < \alpha < 1\).

data: pandas.DataFrame#

The simulation result as a data frame, indexed by the central values of the partitions for which the simulation was run, with the following columns:

COL_MEAN: the mean predictions for the simulated values
COL_SEM: the standard errors of the mean predictions
COL_LOWER_BOUND: the lower bounds of the confidence intervals for the simulation outcomes, based on mean, standard error of the mean, and confidence_level
COL_UPPER_BOUND: the upper bounds of the confidence intervals for the simulation outcomes, based on mean, standard error of the mean, and confidence_level

feature_name: str#: Name of the simulated feature.

output_name: str#: Name of the target for which outputs are simulated.

output_unit: str#: The unit of the simulated outputs (e.g., uplift or class probability).

partitioner: facet.data.partition.Partitioner[T_Values]#: The partitioner used to generate feature values to be simulated.

facet.simulation.UnivariateProbabilitySimulator

facet.simulation.UnivariateTargetSimulator