calculate_TAE: Calculate the total absolute error (TAE) between sample data...

Description Usage Arguments Examples

View source: R/sim_annealing.R

Description

Calculates the total absolute error (TAE) between sample micro data and constraining totals from the matching macro data. Allows for updating of prior TAE instead of re-calculating to improve speed in iterating. The updating feature is particularly helpful for optimizing micro data fitting via simulated annealing (see optimize_microdata).

Usage

1
2
calculate_TAE(sample_data, constraint_list, prior_sample_totals = NULL,
  dropped_obs_totals = NULL, new_obs = NULL)

Arguments

sample_data

A data.frame with attributes matching constraint_list.

constraint_list

A list of constraints. See add_constraint.

prior_sample_totals

An optional list containing attribute counts of a prior sample corresponding to the constraint list. Defaults to NULL.

dropped_obs_totals

An optional list containing attribute counts from the dropped observations in a prior sample. Defaults to NULL.

new_obs

An optional data.frame containing new observations with attributes matching those in sample_data, constraint_list, and prior_sample_totals. Defaults to NULL.

Examples

1
2
3
4
5
6
7
8
## Not run: 
## assumes that you have a micro_synthetic dataset named test_micro and attribute count
## named g respectively 
c_list <- add_constraint(attr_name= "gender", attr_totals= g, micro_data= test_micro,
            constraint_list= c_list)
calculate_TAE(test_micro, c_list)

## End(Not run)

synthACS documentation built on May 30, 2017, 1:51 a.m.