predict.linearRidge does not work if formula goes out of scope
dfrankow opened this issue · 5 comments
Test case:
library(testthat)
tol <- 0.0001
test_that("test linearRidge with formula variable that goes out of scope", {
the_formula <- formula('mpg ~ wt + cyl')
model1 <- lm(the_formula, data = mtcars)
model2 <- linearRidge(the_formula, data = mtcars, lambda = 0)
# suppose the_formula goes away:
rm(the_formula)
expect_equal(coef(model1), coef(model2), tolerance = tol, label = "coefficients")
expect_equal(predict(model1), predict(model2), tolerance = tol, label = "predictions")
})
This is separate from my changes, and also fails on 2.4.
The failure message:
Error: Test failed: 'test linearRidge with formula variable that goes out of scope'
* object 'the_formula' not found
Backtrace:
1. stats::model.frame(formula = the_formula, data = mtcars)
lm handles this, I think by saving the environment somehow. It has tons of specialized functions, including model.matrix.lm and model.frame.lm that set up and move around data.
This issue is important because if a linearRidge model is set up in a function (that returns the model, say), when that function returns, everything in it will go out of scope, and the model won't be able to predict.
Wow, does this pull request already fix the issue? You are so fast 🥇
I have to go to bed now ... hope I can run my tests tomorrow. Today I was more busy with fixing my Mac issues...and was not able to actually test much.
Glad you think so. Thanks for your quick response.
These issues are blocking me, so I've been working on it a bunch. However, work at your own pace. In a pinch I can use my fork, so your release is not blocking me.