coef.LarsPath {HDPenReg} | R Documentation |
Compute coefficients at a given level of penalty
## S3 method for class 'LarsPath' coef(object, index = NULL, mode = c("lambda", "step", "fraction", "norm"), ...)
object |
a LarsParth object |
index |
If mode ="norm", index represents the l1-norm of the coefficients with which we want to predict. If mode="fraction", index represents the ratio (l1-norm of the coefficientswith which we want to predict)/(l1-norm maximal of the LarsPath object). If mode="lambda", index represents the value of the penalty parameter. If mode="step", index represents the numer of the step at which we want coefficients. |
mode |
"fraction" or "norm" or "lambda" or "step". |
... |
other arguments. Not used |
A vector containing the estimated coefficient for index
Quentin Grimonprez
dataset=simul(50,10000,0.4,10,50,matrix(c(0.1,0.8,0.02,0.02),nrow=2)) result=HDlars(dataset$data[1:40,],dataset$response[1:40]) coeff=coef(result,0.3,"fraction")