0.31.0

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: marginaleffects
 Title: Predictions, Comparisons, Slopes, Marginal Means, and Hypothesis Tests
-Version: 0.30.0.7
+Version: 0.31.0
 Authors@R:
     c(person(given = "Vincent",
              family = "Arel-Bundock",

NEWS.md

@@ -1,17 +1,14 @@
 # News {.unnumbered}
 
-## 0.30.0.7
+## 0.31.0
 
-New:
-
-* `inferences(method = "conformal_full")` implements full conformal prediction for regression models. This provides distribution-free prediction intervals by refitting the model on augmented data for each test observation. Unlike split conformal or CV+, full conformal uses all training data without requiring a separate calibration set. It is computationally intensive but provides valid coverage guarantees. For tidymodels workflows, pass the training data explicitly via the `data_train` argument.
-
-Breaking changes:
+Deprecated:
 
 * The `conformal_calibration`, `conformal_train`, and `conformal_test` arguments in `inferences()` are deprecated (but kept for backward compatibility). Use the new `data_calib`, `data_train`, and `data_test`.
 
-## 0.30.1
+New:
 
+* `inferences(method = "conformal_full")` implements full conformal prediction for regression models. This provides distribution-free prediction intervals by refitting the model on augmented data for each test observation. Unlike split conformal or CV+, full conformal uses all training data without requiring a separate calibration set. It is computationally intensive but provides valid coverage guarantees. For tidymodels workflows, pass the training data explicitly via the `data_train` argument.
 * Add support `MASS::lda()` models for estimates but not standard errors. Thanks to @friendly for feature request #1598.
 
 Bugs:

R/inferences.R

@@ -24,6 +24,7 @@
 #' + `rsample`: "perc" or "bca"
 #' + `simulation`: "perc" or "wald"
 #' @param data_train Data frame used to train/fit the model. If `NULL`, `marginaleffects` tries to extract the data from the original model object. Test data are inferred directly from the `newdata` supplied to the originating `marginaleffects` call (e.g., `predictions()`).
+#' @param data_test Data frame make out of sample prediction. Only used for conformal inference. If `NULL`, the data frame supplied to `newdata` in the original `marginaleffects` call is used.
 #' @param data_calib Data frame used for calibration in split conformal prediction.
 #' @param conformal_score String. Warning: The `type` argument in `predictions()` must generate predictions which are on the same scale as the outcome variable. Typically, this means that `type` must be "response" or "probs".
 #'   + "residual_abs" or "residual_sq" for regression tasks (numeric outcome)

R/inferences_conformal_quantile.R

@@ -70,7 +70,7 @@ conformal_quantile <- function(x, data_test, data_train, data_calib, conf_level
     # Step 2: Get quantile predictions on calibration set
     # Use same predictor columns in same order as training
     x_calib <- as.data.frame(data_calib)[, predictor_cols, drop = FALSE]
-    q_calib <- predict(qrf, newdata = x_calib, what = c(alpha / 2, 1 - alpha / 2))
+    q_calib <- stats::predict(qrf, newdata = x_calib, what = c(alpha / 2, 1 - alpha / 2))
     q_calib_low <- q_calib[, 1]
     q_calib_high <- q_calib[, 2]
 
@@ -96,7 +96,7 @@ conformal_quantile <- function(x, data_test, data_train, data_calib, conf_level
     # Step 5: Get quantile predictions on test set
     # Use same predictor columns in same order as training
     x_test <- as.data.frame(data_test)[, predictor_cols, drop = FALSE]
-    q_test <- predict(qrf, newdata = x_test, what = c(alpha / 2, 1 - alpha / 2))
+    q_test <- stats::predict(qrf, newdata = x_test, what = c(alpha / 2, 1 - alpha / 2))
     q_test_low <- q_test[, 1]
     q_test_high <- q_test[, 2]
 

inst/WORDLIST

@@ -7,6 +7,7 @@ ARD
 Aaditya
 Adelie
 Adelies
+Alessandro
 Anastasios
 Angelopoulos
 Arel
@@ -26,6 +27,7 @@ CRC
 Candes
 Carlisle
 Codecov
+Conformalized
 Customizable
 Demetri
 Diniz
@@ -40,6 +42,7 @@ Estimand
 Estimands
 Fabbri
 Foygel
+G'Sell
 GLMMs
 Gam
 Gentoos
@@ -61,6 +64,7 @@ J'VJ
 JAX
 JM
 JVJ
+Jing
 Jørgensen
 Kapre
 Kirill
@@ -97,6 +101,7 @@ Rdatasets
 Rdatatable
 Resul
 Rina
+Rinaldo
 Rmarkdown
 Robb
 Rohan
@@ -117,10 +122,12 @@ Tomz
 Typst
 Uffe
 Umit
+Wasserman
 Westfall
 Wittenberg
 XGBoost
 Xu
+Yaniv
 aje
 al
 arXiv
@@ -288,6 +295,7 @@ thornton
 threenum
 tibbles
 tidiers
+tidymodels
 timpipeseek
 tinytable
 tukey