Skip to contents

[Experimental]

From any result parameter and its corresponding cfp_run_map calculate first-order and total sobol indices using the Azzini (2021) method.

Usage

sobol_calc_indices(Y, effect_cols, id_cols = character(), run_map)

Arguments

Y

A data.frame with the desired effect parameter(s) of the model output, e.g. efflux(). The output should come from a list of model results produced by a call to alternate() with a valid cfp_run_map produced by sobol_run_map().

effect_cols

character vector of the column names in Y for which sobol indices should be calculated, e.g. 'efflux'.

id_cols

character vector of column names in Y specifying grouping variables. Indices are then calculated for each group individually.

run_map

The cfp_run_map used for the calculation of Y produced by a call to sobol_run_map().

Value

A data.frame with the following columns

...

Any id_cols specified

param_id, param, pmap

Parameter identificators from the cfp_run_map used.

effect_param

The parameter for which the effect was calculated.

Vt, Vi, VY

Internal parameters for the indice calculation.

Si

First order sobol indice.

ST

Total order sobol indice.

Details

This implements the approach outlined in Azzini et al (2021).

References

Azzini, Ivano; Mara, Thierry A.; Rosati, Rossana: Comparison of two sets of Monte Carlo estimators of Sobol’ indices, Environmental Modelling & Software, Volume 144, 2021, 105167, ISSN 1364-8152, https://doi.org/10.1016/j.envsoft.2021.105167

See also

Other sobol: sobol_run_map()