nnnusvr: nnnusvr

In mskilab/skitools: Various mskilab R utilties

nnnusvr

Description

Implementation of non-negative linear nu-support vector regression using CPLEX, given length n response vector y and n x m data matrix X returns a length m weight vector w linear SV regresssion result (or length m+1 result if intercept = TRUE is specified).

Additional feature nn (default = TRUE) allows constraining the weight vector to be non-negative

Usage

nnnusvr(X, y, nu = 0.85, C = 1, nn = TRUE, intercept = FALSE)

Arguments

X	numeric matrix of n data points by m features
y	numeric length(n) response vector
nu	scalar numeric between 0 and 1 representing fraction of data to use as support vectors
C	scalar non-negative numeric representing the cost
nn	logical scalar whether to constrain the output weights to be non-negative

Author(s)

Marcin Imielinski

mskilab/skitools documentation built on Aug. 31, 2023, 1:13 p.m.