SensitivityDistribution

class SensitivityDistribution(problem, metaSAAlgorithm, sampleSize, numberOfRepetitions=10, estimator='Saltelli', sampling_method='MonteCarlo')

Checks the distribution of the sensitivity indices estimators.

We comapre the asymptotic distribution with an empirical sample of Sobol’ indices estimates.

We want to check that the distribution computed by the library is OK. When we estimate this distribution depending on the asymptotic estimator, this distribution is Gaussian. Otherwise, it is computed using kernel smoothing based on bootstrap. A first significant point is the mean of the distribution. The second significant point is the variance of the distribution, which is computed using the Delta-method when the asymptotic distribution is used.

The reference distribution is computed based on a sample of Sobol’ indices, which is generated by repetition. Then we compare the “Sample” distribution (from repetition) and the “Computed” distribution from the library. These two distributions should be close.

Methods

compute_sample_indices()

Generate a sample of first order and total order Sobol' indices.

draw([mean_distribution])

Plot the distribution of the estimator and the distribution of the indices.
__init__(problem, metaSAAlgorithm, sampleSize, numberOfRepetitions=10, estimator='Saltelli', sampling_method='MonteCarlo')

Checks the distribution of the Sobol’ estimator.

Parameters:
problemot.SensitivityBenchmarkProblem

The problem.

metaSAAlgorithmSensitivityBenchmarkMetaAlgorithm

A meta-sensitivity algorithm.

sampleSizeint, optional

The sample size.

numberOfRepetitionsint, optional

The number of times the estimation is repeated. The default is 10.

estimatorstr

The estimator. Must be “Saltelli”, “Jansen”, “Martinez”, “MauntzKucherenko”.

sampling_methodstr

The sampling method. Must be “MonteCarlo” or “LHS” or “QMC”.

Returns:
None.
compute_sample_indices()

Generate a sample of first order and total order Sobol’ indices.

Returns:
sampleFirstot.Sample(numberOfRepetitions, dimension)

A sample of first order Sobol’ indices.

sampleTotalTYPE

A sample of total order Sobol’ indices.

distributionFirstot.Distribution

The distribution of the first order Sobol’ indices..

distributionTotalTYPE

The distribution of the total order Sobol’ indices..

draw(mean_distribution=False)

Plot the distribution of the estimator and the distribution of the indices.

Parameters:
mean_distributionTYPE, optional

DESCRIPTION. The default is False.

Returns:
None.