0

I made a simple df as follows:

df<-data.frame(pass=c(0,1,0,0,1,1,1,0,0,1,0,1,1,1,0,0,0,0,0,1),
               math=c(23,46,66,78,77,88,90,99,21,34,56,55,67,67,88,89,90,12,11,34),
               physics=c(87,43,56,78,44,56,90,99,21,32,45,46,46,77,88,90,32,12,34,57),
               bmi=c(23,24,34,21,18,19,26,37,35,21,12,13,41,25,27,28,34,32,21,22))

pass is a response variable for logistic lasso regression. I made train and test set, fit logistic lasso regression. I could get coefficients of logistic lasso regression by using coef(lasso.model).However, I don't know how to get p-values.

#split train and test set
sample <- sample.int(n = nrow(df), size = floor(.7*nrow(df)), replace = F)
train <- df[sample, ]
test  <- df[-sample, ]

library(ROCR)
library(glmnet)
library(caret)

x        <- as.matrix(data.frame(train[,2:4]))
y<-as.matrix(train$pass)

lasso.model <- cv.glmnet(x , y , 
                         family = 'binomial', type.measure = 'auc')


newX <- as.matrix(data.frame(test[,2:4]))

# Apply model to testing dataset
test$lasso.prob <- predict(lasso.model,type="response", 
                         newx = newX, s = 'lambda.min')

coef(lasso.model)
Lee
  • 369
  • 1
  • 6
  • 2
    See here: https://stackoverflow.com/questions/12937331/why-is-it-inadvisable-to-get-statistical-summary-information-for-regression-coef/17725220#17725220. There are not available – Maël Sep 05 '22 at 14:30
  • Thanks for the comment. I 've thought I could get p-values from that package. – Lee Sep 05 '22 at 14:33
  • 2
    @Lee This reading might be interesting for you: https://stats.stackexchange.com/q/45449 – jay.sf Sep 05 '22 at 15:07

0 Answers0