Predictions using glmnet in R

I am trying to model some data using the glmnet package in R. Let’s say I have the following data

training_x <- data.frame(variable1 = c(1, 2, 3, 2, 3),
                         variable2 = c(1, 2, 3, 4, 5))
y <- c(1, 2, 3, 4, 5)

(This is a simplification; my data are much more complicated.) Then I used the following code to create the glmnet model.

x <- as.matrix(training_x)
GLMnet_model_1 <- glmnet(x, y, family="gaussian", alpha=0.755,
                         nlambda=1000, standardize=FALSE, maxit=100000)

I am using standardize=FALSE because my real life data are already standardized. Then I want to make prediction over a new set of data. Let’s say my new data are:

newdata <- as.matrix(data.frame(variable1 = c(2, 2, 1, 3), 
                                variable2 = c(6, 2, 1, 3)))
results <- predict(object=GLMnet_model_1, newx, type="response")

I would expect results to contain 4 elements (predictions of the newdata), but instead it gives me a 4×398 matrix. What am I doing wrong?


You need to specify for which value of lambda you want to predict the response. All you need to do is to call like like e.g.:

results <-predict(GLMnet_model_1, s=0.01, newx, type="response")

Source : Link , Question Author : Benoit_Plante , Answer Author : AlefSin

Leave a Comment