Breadcrumbs

Perturbed Estimates


Summary


Adjusts the sum of a measure so that the ratio of perturbed sum and perturbed mean is equal to the perturbed contributor count.

Notes

This module adjusts the sum of a measure to make it consistent with the perturbed count by multiplying the unperturbed mean of the measure by the perturbed contributor count:

perturbed sum = (unperturbed sum / unperturbed contributor count) x perturbed contributor count

This has the effect of proportionately scaling the sum to be consistent with the change that perturbation has applied to the count.

The "contributor count" used in the calculation is the count of non-null values for that measure. This will be less than the record count if the data has null values. You should therefore take care to ensure that any genuine null values in the data are channelled to the SXV4 as null, rather than as 0.  

This module is not required if you are only using discrete perturbation and do not have any measures available.

Filename

perturbedestimates

Dependencies

Perturbation

Priority

Perturbed estimates uses the results of discrete perturbation in its calculation, so it must always run after the perturbation module. If you are using weighted datasets and/or continuous perturbation, then the modules must run in the following sequence:

  1. Discrete Perturbation.

  2. (Optional, if weighted datasets in use). Average Cell Weight.

  3. (Optional, if continuous/top n perturbation required). Continuous Perturbation.

  4. Perturbed Estimates.

Configuration 

method addmethod <method_id> [{table|mandatory}] [<display_name>]
method <method_id> adddcplugin <plugin_id> perturbedestimates [<priority>]

Example 

method addmethod MethodID 
method MethodID adddcplugin Perturbation perturbation
method MethodID Perturbation addproperty RKEY "TRUE"
method MethodID adddcplugin PerturbedEstimates perturbedestimates

Properties

None