Obtains predictive intervals on a testing dataset based on a conformalCV object from conformal with useCV = TRUE.

# S3 method for conformalCV
predict(object, Xtest, alpha = 0.1, wthigh = 20, wtlow = 0.05, ...)

Arguments

object

an object of class conformalCV; see conformal.

Xtest

testing covariates.

alpha

confidence level.

wthigh

upper truncation level of weights; see Details.

wtlow

lower truncation level of weights; see Details.

...

other arguments

Value

predictive intervals. A data.frame with nrow(Xtest) rows and two columns: "lower" for the lower bound and "upper" for the upper bound.

Details

Given a testing set \(X_1, X_2, \ldots, X_n\) and a weight function \(w(x)\), the weight of the weighted distribution \(p_j = w(X_j) / (w(X_1) + \cdots + w(X_n))\). In cases where some of \(p_j\) are extreme, we truncate \(p_j\) at level wthigh and wtlow to ensure stability. If wthigh = Inf, wtlow = 0, no truncation is being used.