-
Notifications
You must be signed in to change notification settings - Fork 47
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Support topmodels
#1159
base: main
Are you sure you want to change the base?
Support topmodels
#1159
Conversation
@zeileis I have not tried too many things yet, but at first glance what we discussed seems to work well. See above for installation and examples. |
Vincent, really cool, thanks! I did not try many things, yet, but noticed the following. While Also, the Thus, instead of More tomorrow! |
TODO:
|
In general, I like verbosity, but in this particular case I feel that it might be best to be consistent with A lot of users are asking me to implement prediction intervals in Is this something that library("topmodels")
library("marginaleffects")
sim <- function(...) {
N <- 500
dat <- data.frame(x = rnorm(N))
dat$y <- dat$x + rnorm(N)
idx <- sample(1:N, N / 2)
train <- dat[idx, ]
test <- dat[-idx, ]
m <- glm(y ~ x, data = train)
p <- do.call(cbind, list(
procast(m, newdata = test, type = "mean"),
procast(m, newdata = test, type = "quantile", at = .05),
procast(m, newdata = test, type = "quantile", at = .95)
))
p <- setNames(p, c("estimate", "conf.low", "conf.high"))
p$truth <- test$y
coverage <- mean(p$truth < p$conf.high & p$truth > p$conf.low)
return(coverage)
}
mean(sapply(seq_len(100), sim))
> [1] 0.89572 |
Install:
Examples seem to work: