strapvizpy.bootstrap

Module Contents

Functions

bootstrap_distribution(sample, rep, n='auto', estimator='mean', random_seed=None)

Bootstraps a sampling distribution for a sample.

calculate_boot_stats(sample, rep, n='auto', level=0.95, estimator='mean', random_seed=None, pass_dist=False)

Calculates a bootstrapped confidence interval for a sample.

Attributes

SUPPORTED_ESTIMATORS

strapvizpy.bootstrap.SUPPORTED_ESTIMATORS
strapvizpy.bootstrap.bootstrap_distribution(sample, rep, n='auto', estimator='mean', random_seed=None)[source]

Bootstraps a sampling distribution for a sample.

A sampling distribution of rep replicates is generated for the specified estimator`with replacement with a bootstrap sample size of `n.

Parameters

  • sample (list or numpy.ndarray or pandas.core.series.Series) – sample to bootstrap

  • rep (int) – number of replicates of the distribution

  • n (str or int, default="auto") – bootstrap sample size, “auto” specifies using the same size as the sample

  • estimator ({"mean", "median", "var", "sd"}) – sampling distributor’s estimator

  • random_seed (None or int, default=None) – seed for random state

Returns

bootstrapped sampling distribution

Return type

numpy.ndarray

Examples

>>> bootstrap_distribution([1, 2, 3], 3, 3)
array([1.66, 2, 2.66])
strapvizpy.bootstrap.calculate_boot_stats(sample, rep, n='auto', level=0.95, estimator='mean', random_seed=None, pass_dist=False)[source]

Calculates a bootstrapped confidence interval for a sample.

A bootstrapped confidence interval for the desired estimator for the provided sample is calculated for a confidence level level. Other stats and parameters of the distribution and sample are also returned.

Parameters

  • sample (list or numpy.ndarray or pandas.core.series.Series) – sample to bootstrap

  • rep (int) – number of replicates of the distribution

  • n (str or int, default="auto") – bootstrap sample size, “auto” specifies using the same size as the sample

  • level (float, default=0.95) – confidence level

  • estimator ({"mean", "median", "var", "sd"}) – sampling distributor’s estimator

  • random_seed (None or int, default=None) – seed for random state

  • pass_dist (bool, default = "False") – return the bootstrapped sample distribution - False or True

Returns

Dictionary containing lower and upper bootstrapped confidence interval for the desired estimator, along with the given estimator. Also

Return type

dictionary

Examples

>>> calculate_boot_stats([1, 2, 3, 4], 1000, level=0.95, random_seed=123)
{'lower': 1.5,
'upper': 3.5,
'sample_mean': 2.5,
'std_err': 0.5414773771820943,
'level': 0.95,
'sample size': 4,
'n': 'auto',
'rep': 1000,
'estimator': 'mean'}