lpca - WeAreCAS

Description

The lpca action reduces the dimensionality of nominal variables by using logistic principal component analysis (LPCA). LPCA is a generalization of principal component analysis that is adapted for categorical data. It models the logits of the probabilities of the levels of the nominal variables as a linear combination of a small number of principal components.

nominalVarsDimReduction.lpca <result=results> <status=rc> / dimensions=integer, display={...}, epsilon=double, freq="variable-name", id={"variable-name-1" <, "variable-name-2", ...>}, inputs={{...} <, {...}, ...>}, m=double, maxIter=integer, nominals={{...} <, {...}, ...>}, output={...}, outputTables={...}, prefix="string", saveState={...}, table={...};

Settings

Parameter	Description
dimensions	Specifies the number of reduced variables.
display	Specifies a list of results tables to send to the client for display.
epsilon	Specifies the tolerance to use in determining the convergence of the iterative algorithm.
freq	Specifies the frequency variable.
id	Specifies the variables to use as record identifiers and to transfer to the output table that you specify in the output parameter.
inputs	Specifies the variables to use in the analysis.
m	Specifies a finite positive value to approximate the logit function's infinite limits: logit(1) is approximated by m, and logit(0) is approximated by -m.
maxIter	Specifies the maximum number of iterations for the iterative algorithm.
nominals	Specifies the nominal variables to use in the training.
output	Specifies the output data table that contains the values of the reduced variables for the training nominal data.
outputTables	Lists the names of results tables to save as CAS tables on the server.
prefix	Specifies a prefix to apply to the names of the reduced variables.
saveState	Specifies the output data table in which to save the dimensionality reduction model of the nominal variables for future scoring.
table	Specifies the input table.

Examples

FAQ

What is the primary function of the lpca action?

Which parameter is required to specify the input data table for the lpca action?

How do you specify the number of reduced variables to be created?

What is the purpose of the 'nominals' parameter?

How can I save the trained LPCA model for future scoring?

What does the 'output' parameter do?

What is the role of the 'prefix' parameter?