ignore/problems.R

In bssherwood/rqpen: Penalized Quantile Regression

#I think the following is problematic because there is a kitchen quality with only one value, however should be able to provide a better error message
library(rqPen)
library(AmesHousing)
set.seed(1)
ames <- make_ames()
x_g <- model.matrix(~ Lot_Shape+Garage_Type+Full_Bath+Fireplaces+Kitchen_Qual - 1,ames)
y_g <- log(ames$Sale_Price)
g <-  c(rep(1,4),rep(2,6),3,4,rep(5,4))
r2 <- rq.group.pen.cv(x_g,y_g,groups=g)
r3 <- rq.group.pen.cv(x_g,y_g,groups=g,penalty="gSCAD",norm=1)
r4 <- rq.group.pen.cv(x_g,y_g,groups=g,penalty="gSCAD")