objective {npmr}R Documentation

NPMR objective function

Description

Return the objective function of the data and the fitted parameters for nuclear penalized multinomial regression. The objective is the sum of the negative log-likelihood and the product of the regularization parameter and nuclear norm of the fitted regression coefficient matrix. Intended for internal use only.

Usage

objective(B, b, X, Y, lambda)

Arguments

B

fitted regression coefficient matrix

b

fitted regression intercept vector

X

covariate matrix

Y

multinomial response matrix

lambda

regularization parameter (maybe be a vector of values)

Value

a vector of objective values for the NPMR optimization problem, one for each value of lambda

Author(s)

Scott Powers, Trevor Hastie, Rob Tibshirani

References

Scott Powers, Trevor Hastie and Rob Tibshirani (2016). “Nuclear penalized multinomial regression with an application to predicting at bat outcomes in baseball.” In prep.

See Also

logL, nuclear, objectiveFast


[Package npmr version 1.3.1 Index]