Solving the nthread problem?

NEWS.md

@@ -2,15 +2,16 @@
 
 ## User-visible changes
 
-- `mshapviz()` objects can now be rowbinded via `rbind()` or `+`. Implemented by @jmaspons in [#110](https://github.com/ModelOriented/shapviz/pull/110).
+- `mshapviz()` objects can now be rowbinded via `rbind()` or `+`. Implemented by [@jmaspons](https://github.com/jmaspons) in [#110](https://github.com/ModelOriented/shapviz/pull/110).
 - `mshapviz()` is more strict when combining multiple "shapviz" objects. These now need to have identical column names, see [#114](https://github.com/ModelOriented/shapviz/pull/114).
 
 ## Small changes
 
 - `print.shapviz()` now shows top two rows of SHAP matrix.
 - Re-activate all unit tests.
+- Setting `nthread = 1` in all calls to `xgb.DMatrix()` as suggested by [@jmaspons](https://github.com/jmaspons) in [issue #109](https://github.com/ModelOriented/shapviz/issues/109).
 - Added "How to contribute" to README.
-- `permshap()` connector is now part of {kerneshap}.
+- `permshap()` connector is now part of {kerneshap} [#122](https://github.com/ModelOriented/shapviz/pull/122).
 
 ## Bug fixes
 

R/shapviz.R

@@ -114,7 +114,7 @@ shapviz.matrix = function(object, X, baseline = 0, collapse = NULL,
 #' \dontrun{
 #' # XGBoost models
 #' X_pred <- data.matrix(iris[, -1])
-#' dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1], nthread = 1)
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #'
 #' # Will use numeric matrix "X_pred" as feature matrix
@@ -132,8 +132,10 @@ shapviz.matrix = function(object, X, baseline = 0, collapse = NULL,
 #' # Multiclass setting
 #' params <- list(objective = "multi:softprob", num_class = 3)
 #' X_pred <- data.matrix(iris[, -5])
-#' dtrain <- xgboost::xgb.DMatrix(X_pred, label = as.integer(iris[, 5]) - 1)
-#' fit <- xgboost::xgb.train(params = params, data = dtrain, nrounds = 10)
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   X_pred, label = as.integer(iris[, 5]) - 1, nthread = 1
+#' )
+#' fit <- xgboost::xgb.train(params = params, data = dtrain, nrounds = 10, nthread = 1)
 #'
 #' # Select specific class
 #' x <- shapviz(fit, X_pred = X_pred, which_class = 3)
@@ -145,7 +147,7 @@ shapviz.matrix = function(object, X, baseline = 0, collapse = NULL,
 #'
 #' # What if we would have one-hot-encoded values and want to explain the original column?
 #' X_pred <- stats::model.matrix(~ . -1, iris[, -1])
-#' dtrain <- xgboost::xgb.DMatrix(X_pred, label = as.integer(iris[, 1]))
+#' dtrain <- xgboost::xgb.DMatrix(X_pred, label = as.integer(iris[, 1]), nthread = 1)
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #' x <- shapviz(
 #'   fit,

R/sv_dependence.R

@@ -35,7 +35,9 @@
 #' @returns An object of class "ggplot" (or "patchwork") representing a dependence plot.
 #' @examples
 #' \dontrun{
-#' dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1]), label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   data.matrix(iris[, -1]), label = iris[, 1], nthread = 1
+#' )
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #' x <- shapviz(fit, X_pred = dtrain, X = iris)
 #' sv_dependence(x, "Petal.Length")

R/sv_dependence2D.R

@@ -30,7 +30,9 @@
 #' @returns An object of class "ggplot" (or "patchwork") representing a dependence plot.
 #' @examples
 #' \dontrun{
-#' dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1]), label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   data.matrix(iris[, -1]), label = iris[, 1], nthread = 1
+#' )
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #' sv <- shapviz(fit, X_pred = dtrain, X = iris)
 #' sv_dependence2D(sv, x = "Petal.Length", y = "Species")

R/sv_force.R

@@ -12,7 +12,9 @@
 #' @returns An object of class "ggplot" (or "patchwork") representing a force plot.
 #' @examples
 #' \dontrun{
-#' dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1]), label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   data.matrix(iris[, -1]), label = iris[, 1], nthread = 1
+#' )
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 20, nthread = 1)
 #' x <- shapviz(fit, X_pred = dtrain, X = iris[, -1])
 #' sv_force(x)

R/sv_importance.R

@@ -48,7 +48,7 @@
 #' @examples
 #' \dontrun{
 #' X_train <- data.matrix(iris[, -1])
-#' dtrain <- xgboost::xgb.DMatrix(X_train, label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(X_train, label = iris[, 1], nthread = 1)
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #' x <- shapviz(fit, X_pred = X_train)
 #' sv_importance(x)

R/sv_interaction.R

@@ -19,7 +19,9 @@
 #'   absolute SHAP values (or a list of such matrices in case of "mshapviz" object).
 #' @examples
 #' \dontrun{
-#' dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1]), label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   data.matrix(iris[, -1]), label = iris[, 1], nthread = 1
+#' )
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 10, nthread = 1)
 #' x <- shapviz(fit, X_pred = dtrain, X = iris, interactions = TRUE)
 #' sv_interaction(x, kind = "no")

R/sv_waterfall.R

@@ -36,7 +36,9 @@
 #' @returns An object of class "ggplot" (or "patchwork") representing a waterfall plot.
 #' @examples
 #' \dontrun{
-#' dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1]), label = iris[, 1])
+#' dtrain <- xgboost::xgb.DMatrix(
+#'   data.matrix(iris[, -1]), label = iris[, 1], nthread = 1
+#' )
 #' fit <- xgboost::xgb.train(data = dtrain, nrounds = 20, nthread = 1)
 #' x <- shapviz(fit, X_pred = dtrain, X = iris[, -1])
 #' sv_waterfall(x)

tests/testthat/test-collapse_shap.R

@@ -87,7 +87,7 @@ test_that("collapse_shap works for SHAP interactions and two collapses (result i
 # # Real data example
 form <- Sepal.Length ~ Sepal.Width + Species - 1
 iris_dummy <- model.matrix(form, data = iris)
-dtrain <- xgboost::xgb.DMatrix(iris_dummy, label = iris[, 1L])
+dtrain <- xgboost::xgb.DMatrix(iris_dummy, label = iris[, 1L], nthread = 1)
 fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
 coll <- list(Species = paste0("Species", levels(iris$Species)))
 

tests/testthat/test-interface.R

@@ -259,7 +259,7 @@ test_that("combining incompatible shapviz objects fails", {
 
 # Multiclass with XGBoost
 X_pred <- data.matrix(iris[, -5L])
-dtrain <- xgboost::xgb.DMatrix(X_pred, label = as.integer(iris[, 5L]) - 1L)
+dtrain <- xgboost::xgb.DMatrix(X_pred, label = as.integer(iris[, 5L]) - 1L, nthread = 1)
 fit <- xgboost::xgb.train(
   params = list(nthread = 1L),
   data = dtrain,

tests/testthat/test-plots-mshapviz.R

@@ -1,4 +1,6 @@
-dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1L]), label = iris[, 1L])
+dtrain <- xgboost::xgb.DMatrix(
+  data.matrix(iris[, -1L]), label = iris[, 1L], nthread = 1
+)
 fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
 x <- shapviz(fit, X_pred = dtrain, X = iris[, -1L])
 x <- c(m1 = x, m2 = x)
@@ -83,7 +85,7 @@ test_that("Interaction plots provide patchwork object", {
 # Non-standard name
 ir <- iris
 ir["strange name"] <- ir$Sepal.Width * ir$Petal.Length
-dtrain <- xgboost::xgb.DMatrix(data.matrix(ir[, -1L]), label = ir[, 1L])
+dtrain <- xgboost::xgb.DMatrix(data.matrix(ir[, -1L]), label = ir[, 1L], nthread = 1)
 fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
 x <- shapviz(fit, X_pred = dtrain, X = ir[, -1L])
 x <- c(m1 = x, m2 = x)
@@ -108,7 +110,7 @@ test_that("plots work for non-syntactic column names", {
 
 test_that("sv_importance() and sv_interaction() and kind = 'no' gives matrix", {
   X_pred <- data.matrix(iris[, -1L])
-  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L])
+  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L], nthread = 1)
   fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
   x <- shapviz(fit, X_pred = X_pred, interactions = TRUE)
   x <- c(m1 = x, m2 = x)
@@ -122,7 +124,7 @@ test_that("sv_importance() and sv_interaction() and kind = 'no' gives matrix", {
 
 test_that("sv_dependence() does not work with multiple v", {
   X_pred <- data.matrix(iris[, -1L])
-  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L])
+  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L], nthread = 1)
   fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
   x <- c(m1 = shapviz(fit, X_pred = X_pred), m2 = shapviz(fit, X_pred = X_pred))
   expect_error(sv_dependence(x, v = c("Species", "Sepal.Width")))

tests/testthat/test-plots-shapviz.R

@@ -1,4 +1,6 @@
-dtrain <- xgboost::xgb.DMatrix(data.matrix(iris[, -1L]), label = iris[, 1L])
+dtrain <- xgboost::xgb.DMatrix(
+  data.matrix(iris[, -1L]), label = iris[, 1L], nthread = 1
+)
 fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
 x <- shapviz(fit, X_pred = dtrain, X = iris[, -1L])
 
@@ -124,7 +126,7 @@ test_that("Interaction plots provide ggplot object", {
 # Non-standard name
 ir <- iris
 ir["strange name"] <- ir$Sepal.Width * ir$Petal.Length
-dtrain <- xgboost::xgb.DMatrix(data.matrix(ir[, -1L]), label = ir[, 1L])
+dtrain <- xgboost::xgb.DMatrix(data.matrix(ir[, -1L]), label = ir[, 1L], nthread = 1)
 fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
 x <- shapviz(fit, X_pred = dtrain, X = ir[, -1L])
 
@@ -160,7 +162,7 @@ test_that("there are no default sv_*() methods", {
 
 test_that("sv_importance() and sv_interaction() and kind = 'no' gives numeric output", {
   X_pred <- data.matrix(iris[, -1L])
-  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L])
+  dtrain <- xgboost::xgb.DMatrix(X_pred, label = iris[, 1L], nthread = 1)
   fit <- xgboost::xgb.train(params = list(nthread = 1L), data = dtrain, nrounds = 1L)
   x <- shapviz(fit, X_pred = X_pred, interactions = TRUE)
 

vignettes/basic_use.Rmd

@@ -88,7 +88,7 @@ diamonds[, ord] <- lapply(diamonds[, ord], factor, ordered = FALSE)
 
 # Fit XGBoost model
 x <- c("carat", "clarity", "cut", "color")
-dtrain <- xgb.DMatrix(data.matrix(diamonds[x]), label = diamonds$price)
+dtrain <- xgb.DMatrix(data.matrix(diamonds[x]), label = diamonds$price, nthread = 1)
 fit <- xgb.train(
   params = list(learning_rate = 0.1, nthread = 1), data = dtrain, nrounds = 65
 )

vignettes/geographic.Rmd

@@ -71,8 +71,8 @@ y_train <- log(miami$SALE_PRC[ix])
 y_valid <- log(miami$SALE_PRC[-ix])
 
 # Fit XGBoost model with early stopping
-dtrain <- xgb.DMatrix(X_train, label = y_train)
-dvalid <- xgb.DMatrix(X_valid, label = y_valid)
+dtrain <- xgb.DMatrix(X_train, label = y_train, nthread = 1)
+dvalid <- xgb.DMatrix(X_valid, label = y_valid, nthread = 1)
 
 params <- list(
   learning_rate = 0.2, objective = "reg:squarederror", max_depth = 5, nthread = 1
@@ -127,8 +127,8 @@ x2 <- c(x, more_geo)
 X_train2 <- data.matrix(miami[ix, x2])
 X_valid2 <- data.matrix(miami[-ix, x2])
 
-dtrain2 <- xgb.DMatrix(X_train2, label = y_train)
-dvalid2 <- xgb.DMatrix(X_valid2, label = y_valid)
+dtrain2 <- xgb.DMatrix(X_train2, label = y_train, nthread = 1)
+dvalid2 <- xgb.DMatrix(X_valid2, label = y_valid, nthread = 1)
 
 # Build interaction constraint vector
 ic <- c(

vignettes/multiple_output.Rmd

@@ -47,7 +47,7 @@ library(xgboost)
 
 params <- list(objective = "multi:softprob", num_class = 3, nthread = 1)
 X_pred <- data.matrix(iris[, -5])
-dtrain <- xgb.DMatrix(X_pred, label = as.integer(iris[, 5]) - 1)
+dtrain <- xgb.DMatrix(X_pred, label = as.integer(iris[, 5]) - 1, nthread = 1)
 fit <- xgb.train(params = params, data = dtrain, nrounds = 50)
 
 # Create "mshapviz" object (logit scale)
@@ -125,7 +125,7 @@ library(patchwork)
 library(xgboost)
 
 X_pred <- data.matrix(iris[, -1])
-dtrain <- xgb.DMatrix(X_pred, label = iris[, 1])
+dtrain <- xgb.DMatrix(X_pred, label = iris[, 1], nthread = 1)
 fit_xgb <- xgb.train(params = list(nthread = 1), data = dtrain, nrounds = 50)
 
 # Create "mshapviz" object