Revision of mxRowObjective-help from Thu, 10/27/2011 - 12:23

The revisions let you track differences between multiple versions of a post.

Description:This function creates a new MxRowObjective object.

Usage: mxRowObjective(rowAlgebra, reduceAlgebra, dimnames, rowResults = "rowResults", filteredDataRow = "filteredDataRow", existenceVector = "existenceVector")

Arguments:

rowAlgebra: A character string indicating the name of the algebra to be evaluated row-wise.

reduceAlgebra: A character string indicating the name of the algebra that collapses the row results into a single number which is then optimized.

dimnames: A character vector of names corresponding to columns be extracted from the data set.

rowResults: The name of the auto-generated "rowResults" matrix. See details.

filteredDataRow: The name of the auto-generated "filteredDataRow" matrix. See details.

existenceVector: The name of the auto-generated "existenceVector" matrix. See details.

Details: Objective functions are functions for which free parameter values are chosen such that the value of the objective function is minimized. The mxRowObjective function evaluates a user-defined MxAlgebra object called the ‘rowAlgebra’ in a row-wise fashion. It then stores results of the row-wise evaluation in another MxAlgebra object called the ‘rowResults’. Finally, the mxRowObjective function collapses the row results into a single number which is then used for optimization. The MxAlgebra object named by the ‘reduceAlgebra’ collapses the row results into a single number.

The ‘filteredDataRow’ is populated in a row-by-row fashion with all the non-missing data from the current row. You cannot assume that the length of the filteredDataRow matrix remains constant (unless you have no missing data). The ‘existenceVector’ is populated in a row-by-row fashion with a value of 1.0 in column j if a non-missing value is present in the data set in column j, and a value of 0.0 otherwise. Use the functions omxSelectRows, omxSelectCols, and omxSelectRowsAndCols to shrink other matrices so that their dimensions will be conformable to the size of ‘filteredDataRow’.

Value: Returns a new MxRowObjective object. MxRowObjective objects should be included with models with referenced MxAlgebra objects.

Examples:

# Model that adds two data columns row-wise, then sums that column
# Notice no optimization is performed here.
xdat <- data.frame(a=rnorm(10), b=1:10) # Make data set
amod <- mxModel(
	            mxData(observed=xdat, type='raw'),
	            mxAlgebra(sum(filteredDataRow), name = 'rowAlgebra'),
	            mxAlgebra(sum(rowResults), name = 'reduceAlgebra'),
	            mxRowObjective(
	                           rowAlgebra='rowAlgebra',
	                           reduceAlgebra='reduceAlgebra',
	                           dimnames=c('a','b'))
	           )
 
# Model that find the parameter that minimizes the sum of the
#  squared difference between the parameter and a data row.
bmod <- mxModel(
	            mxData(observed=xdat, type='raw'),
	            mxMatrix(values=.75, ncol=1, nrow=1, free=TRUE, name='B'),
	            mxAlgebra((filteredDataRow - B) ^ 2, name='rowAlgebra'),
	            mxAlgebra(sum(rowResults), name='reduceAlgebra'),
	            mxRowObjective(
	                           rowAlgebra='rowAlgebra',
	                           reduceAlgebra='reduceAlgebra',
	                           dimnames=c('a'))
	           )