Lasso regression with 10-fold cross validation

[Carlos2]

I am trying to implement a Lasso regression in the College data set from ISLR, with 10 fold cross-validation. The variable that I am trying to predict is a variable named cost, created by adding up the out of state tuition and the living costs. However, I do not understand the error R is throwing. Could you help me understand what is wrong?

library(ggplot2)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(tidyverse)
library(tidymodels)
#> -- Attaching packages -------------------------------------------------------------- tidymodels 0.1.0 --
#> v broom     0.5.5     v rsample   0.0.5
#> v dials     0.0.4     v tune      0.0.1
#> v infer     0.5.1     v workflows 0.1.0
#> v parsnip   0.0.5     v yardstick 0.0.5
#> v recipes   0.1.9
#> -- Conflicts ----------------------------------------------------------------- tidymodels_conflicts() --
#> x scales::discard()   masks purrr::discard()
#> x dplyr::filter()     masks stats::filter()
#> x recipes::fixed()    masks stringr::fixed()
#> x dplyr::lag()        masks stats::lag()
#> x dials::margin()     masks ggplot2::margin()
#> x yardstick::spec()   masks readr::spec()
#> x recipes::step()     masks stats::step()
#> x recipes::yj_trans() masks scales::yj_trans()
library(glmnet)
#> Loading required package: Matrix
#> 
#> Attaching package: 'Matrix'
#> The following objects are masked from 'package:tidyr':
#> 
#>     expand, pack, unpack
#> Loaded glmnet 3.0-2
library(reprex)
library(ISLR)
data("College")
College <- as_tibble(College) %>% 
  mutate(cost = Outstate + Room.Board) %>% 
  select(-c(Outstate,Room.Board))
College <- na.omit(College)
College<-College %>%
  mutate(Private=if_else(Private == "yes", 1L, 0L))

set.seed(123)
cost_split <- College %>% initial_split(prop = 0.5)
cost_train <- training(cost_split)
cost_test <- testing(cost_split)
x = as.matrix(cost_train[,1:14]) #predictors
y = as.matrix(cost_train[,17]) #goal

lambdas_to_try <- 10^seq(from=1, to=1000)
# Setting alpha = 1 implements lasso regression
lasso_cv <- cv.glmnet(x, y, alpha = 1, lambda = lambdas_to_try,
                      standardize = TRUE, nfolds = 10)
#> Error in elnet(x, is.sparse, ix, jx, y, weights, offset, type.gaussian, : NA/NaN/Inf in foreign function call (arg 15)

[nirgrahamuk]

you are asking it to calculate on Infinity values for lambda.
consider this:

lambdas_to_try <- 10^seq(from=1, to=1000)
tail(lambdas_to_try)
lambdas_to_try <- lambdas_to_try[lambdas_to_try!=Inf]
tail(lambdas_to_try)

after the 308th power on my system the huge number becomes Inf

[Carlos2]

Thank you so much! Now it makes sense. Now I am using this and it is working:

    lambdas_to_try <- 10^seq(-3, 5, length.out = 100)

However, if I want to consider values for lambda up to 1,000, how can I do that?

Thanks again for your help!

[nirgrahamuk]

you can consider 10's of thousands of lambdas, its just you cant have the exponents going so high that the numbers blow up.
I don't know your domain but this is a possibility

lambdas_to_try <- seq(10^-3, 10^5, length.out = 10000)

or just add more to your new setup

 lambdas_to_try <- 10^seq(-3, 5, length.out = 10000)