Merge pull request #127 from ModelOriented/simplify-mlr3

Greatly simplify mlr3 workflow

Author: mayer79

.github/workflows/test-coverage.yaml

@@ -33,11 +33,8 @@ jobs:
             clean = FALSE,
             install_path = file.path(Sys.getenv("RUNNER_TEMP"), "package"),
             function_exclusions = c(
-              "kernelshap\\.Learner",
               "kernelshap\\.ranger",
-              "permshap\\.Learner",
-              "permshap\\.ranger",
-              "mlr3_pred_fun"
+              "permshap\\.ranger"
             )
           )
         shell: Rscript {0}

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: kernelshap
 Title: Kernel SHAP
-Version: 0.4.1
+Version: 0.4.2
 Authors@R: c(
     person("Michael", "Mayer", , "mayermichael79@gmail.com", role = c("aut", "cre")),
     person("David", "Watson", , "david.s.watson11@gmail.com", role = "aut"),

NAMESPACE

@@ -1,9 +1,7 @@
 # Generated by roxygen2: do not edit by hand
 
-S3method(kernelshap,Learner)
 S3method(kernelshap,default)
 S3method(kernelshap,ranger)
-S3method(permshap,Learner)
 S3method(permshap,default)
 S3method(permshap,ranger)
 S3method(print,kernelshap)

NEWS.md

@@ -1,3 +1,14 @@
+# kernelshap 0.4.2
+
+## API
+
+- {mlr3}: Non-probabilistic classification now works. 
+- {mlr3}: For *probabilistic* classification, you now have to pass `predict_type = "prob"`.
+
+## Documentation
+
+- The README has received an {mlr3} and {caret} example.
+
 # kernelshap 0.4.1
 
 ## Performance improvements

R/kernelshap.R

@@ -355,36 +355,3 @@ kernelshap.ranger <- function(object, X, bg_X,
   )
 }
 
-#' @describeIn kernelshap Kernel SHAP method for "mlr3" models, see Readme for an example.
-#' @export
-kernelshap.Learner <- function(object, X, bg_X,
-                               pred_fun = NULL,
-                               feature_names = colnames(X),
-                               bg_w = NULL, exact = length(feature_names) <= 8L,
-                               hybrid_degree = 1L + length(feature_names) %in% 4:16,
-                               paired_sampling = TRUE,
-                               m = 2L * length(feature_names) * (1L + 3L * (hybrid_degree == 0L)),
-                               tol = 0.005, max_iter = 100L, parallel = FALSE,
-                               parallel_args = NULL, verbose = TRUE, ...) {
-  if (is.null(pred_fun)) {
-    pred_fun <- mlr3_pred_fun(object, X = X)
-  }
-  kernelshap.default(
-    object = object,
-    X = X,
-    bg_X = bg_X,
-    pred_fun = pred_fun,
-    feature_names = feature_names,
-    bg_w = bg_w,
-    exact = exact,
-    hybrid_degree = hybrid_degree,
-    paired_sampling = paired_sampling,
-    m = m,
-    tol = tol,
-    max_iter = max_iter,
-    parallel = parallel,
-    parallel_args = parallel_args,
-    verbose = verbose,
-    ...
-  )
-}

R/permshap.R

@@ -168,26 +168,3 @@ permshap.ranger <- function(object, X, bg_X,
   )
 }
 
-#' @describeIn permshap Permutation SHAP method for "mlr3" models, see Readme for an example.
-#' @export
-permshap.Learner <- function(object, X, bg_X,
-                             pred_fun = NULL,
-                             feature_names = colnames(X),
-                             bg_w = NULL, parallel = FALSE, parallel_args = NULL,
-                             verbose = TRUE, ...) {
-  if (is.null(pred_fun)) {
-    pred_fun <- mlr3_pred_fun(object, X = X)
-  }
-  permshap.default(
-    object = object,
-    X = X,
-    bg_X = bg_X,
-    pred_fun = pred_fun,
-    feature_names = feature_names,
-    bg_w = bg_w,
-    parallel = parallel,
-    parallel_args = parallel_args,
-    verbose = verbose,
-    ...
-  )
-}

R/utils.R

@@ -380,26 +380,3 @@ prep_w <- function(w, bg_n) {
   if (!is.double(w)) as.double(w) else w
 }
 
-#' mlr3 Helper
-#' 
-#' Returns the prediction function of a mlr3 Learner.
-#' 
-#' @noRd
-#' @keywords internal
-#' 
-#' @param object Learner object.
-#' @param X Dataframe like object.
-#' 
-#' @returns A function.
-mlr3_pred_fun <- function(object, X) {
-  if ("classif" %in% object$task_type) {
-    # Check if probabilities are available
-    test_pred <- object$predict_newdata(utils::head(X))
-    if ("prob" %in% test_pred$predict_types) {
-      return(function(m, X) m$predict_newdata(X)$prob)
-    } else {
-      stop("Set lrn(..., predict_type = 'prob') to allow for probabilistic classification.")
-    }
-  }
-  function(m, X) m$predict_newdata(X)$response
-}

README.md

@@ -249,9 +249,11 @@ ps_class <- permshap(fit_class, X = iris[, -5], bg_X = iris)
 
 ![](man/figures/README-prob-dep.svg)
 
-### Tidymodels
+### Meta-learners
 
-Meta-learning packages like {tidymodels}, {caret} or {mlr3} are straight-forward to use. The following example additionally shows that the `...` argument of `permshap()` and `kernelshap()` is passed to `predict()`.
+Meta-learning packages like {tidymodels}, {caret} or {mlr3} are straightforward to use. The following examples additionally shows that the `...` arguments of `permshap()` and `kernelshap()` are passed to `predict()`.
+
+#### Tidymodels
 
 ```r
 library(kernelshap)
@@ -285,6 +287,56 @@ $.pred_setosa
 [2,]   0.02628333 0.001315556    0.3683833   0.2706111
 ```
 
+#### caret
+
+```r
+library(kernelshap)
+library(caret)
+
+fit <- train(
+  Sepal.Length ~ ., 
+  data = iris, 
+  method = "lm", 
+  tuneGrid = data.frame(intercept = TRUE),
+  trControl = trainControl(method = "none")
+)
+
+ps <- permshap(fit, iris[-1], bg_X = iris)
+```
+
+#### mlr3
+
+```r
+library(kernelshap)
+library(mlr3)
+library(mlr3learners)
+
+set.seed(1)
+
+task_classif <- TaskClassif$new(id = "1", backend = iris, target = "Species")
+learner_classif <- lrn("classif.rpart", predict_type = "prob")
+learner_classif$train(task_classif)
+
+predict(learner_classif, head(iris))  # setosa setosa        # Classes
+predict(learner_classif, head(iris), predict_type = "prob")  # Probs per class
+
+x <- learner_classif$selected_features()
+
+# For *probabilistic* classification, pass predict_type = "prob" to mlr3's predict()
+ps <- permshap(
+  learner_classif, X = iris, bg_X = iris, feature_names = x, predict_type = "prob"
+)
+ps
+# $setosa
+#      Petal.Length Petal.Width
+# [1,]    0.6666667           0
+# [2,]    0.6666667           0
+
+# Non-probabilistic classification uses auto-OHE internally
+ps <- permshap(learner_classif, X = iris, bg_X = iris, feature_names = x)
+ps
+```
+
 ## References
 
 [1] Erik Štrumbelj and Igor Kononenko. Explaining prediction models and individual predictions with feature contributions. Knowledge and Information Systems 41, 2014.