xgb.plot.shap: SHAP contribution dependency plots

In xgboost: Extreme Gradient Boosting

SHAP contribution dependency plots

Description

Visualizing the SHAP feature contribution to prediction dependencies on feature value.

Usage

xgb.plot.shap(
  data,
  shap_contrib = NULL,
  features = NULL,
  top_n = 1,
  model = NULL,
  trees = NULL,
  target_class = NULL,
  approxcontrib = FALSE,
  subsample = NULL,
  n_col = 1,
  col = rgb(0, 0, 1, 0.2),
  pch = ".",
  discrete_n_uniq = 5,
  discrete_jitter = 0.01,
  ylab = "SHAP",
  plot_NA = TRUE,
  col_NA = rgb(0.7, 0, 1, 0.6),
  pch_NA = ".",
  pos_NA = 1.07,
  plot_loess = TRUE,
  col_loess = 2,
  span_loess = 0.5,
  which = c("1d", "2d"),
  plot = TRUE,
  ...
)

Arguments

data	data as a matrix or dgCMatrix.
shap_contrib	a matrix of SHAP contributions that was computed earlier for the above data. When it is NULL, it is computed internally using model and data.
features	a vector of either column indices or of feature names to plot. When it is NULL, feature importance is calculated, and top_n high ranked features are taken.
top_n	when features is NULL, top_n [1, 100] most important features in a model are taken.
model	an xgb.Booster model. It has to be provided when either shap_contrib or features is missing.
trees	passed to xgb.importance when features = NULL.
target_class	is only relevant for multiclass models. When it is set to a 0-based class index, only SHAP contributions for that specific class are used. If it is not set, SHAP importances are averaged over all classes.
approxcontrib	passed to predict.xgb.Booster when shap_contrib = NULL.
subsample	a random fraction of data points to use for plotting. When it is NULL, it is set so that up to 100K data points are used.
n_col	a number of columns in a grid of plots.
col	color of the scatterplot markers.
pch	scatterplot marker.
discrete_n_uniq	a maximal number of unique values in a feature to consider it as discrete.
discrete_jitter	an amount parameter of jitter added to discrete features' positions.
ylab	a y-axis label in 1D plots.
plot_NA	whether the contributions of cases with missing values should also be plotted.
col_NA	a color of marker for missing value contributions.
pch_NA	a marker type for NA values.
pos_NA	a relative position of the x-location where NA values are shown: min(x) + (max(x) - min(x)) * pos_NA.
plot_loess	whether to plot loess-smoothed curves. The smoothing is only done for features with more than 5 distinct values.
col_loess	a color to use for the loess curves.
span_loess	the span parameter in loess's call.
which	whether to do univariate or bivariate plotting. NOTE: only 1D is implemented so far.
plot	whether a plot should be drawn. If FALSE, only a list of matrices is returned.
...	other parameters passed to plot.

Details

These scatterplots represent how SHAP feature contributions depend of feature values. The similarity to partial dependency plots is that they also give an idea for how feature values affect predictions. However, in partial dependency plots, we usually see marginal dependencies of model prediction on feature value, while SHAP contribution dependency plots display the estimated contributions of a feature to model prediction for each individual case.

When plot_loess = TRUE is set, feature values are rounded to 3 significant digits and weighted LOESS is computed and plotted, where weights are the numbers of data points at each rounded value.

Note: SHAP contributions are shown on the scale of model margin. E.g., for a logistic binomial objective, the margin is prediction before a sigmoidal transform into probability-like values. Also, since SHAP stands for "SHapley Additive exPlanation" (model prediction = sum of SHAP contributions for all features + bias), depending on the objective used, transforming SHAP contributions for a feature from the marginal to the prediction space is not necessarily a meaningful thing to do.

Value

In addition to producing plots (when plot=TRUE), it silently returns a list of two matrices:

• data the values of selected features;

• shap_contrib the contributions of selected features.

References

Scott M. Lundberg, Su-In Lee, "A Unified Approach to Interpreting Model Predictions", NIPS Proceedings 2017, https://arxiv.org/abs/1705.07874

Scott M. Lundberg, Su-In Lee, "Consistent feature attribution for tree ensembles", https://arxiv.org/abs/1706.06060

Examples

data(agaricus.train, package='xgboost')
data(agaricus.test, package='xgboost')

## Keep the number of threads to 1 for examples
nthread <- 1
data.table::setDTthreads(nthread)
nrounds <- 20

bst <- xgboost(agaricus.train$data, agaricus.train$label, nrounds = nrounds,
               eta = 0.1, max_depth = 3, subsample = .5,
               method = "hist", objective = "binary:logistic", nthread = nthread, verbose = 0)

xgb.plot.shap(agaricus.test$data, model = bst, features = "odor=none")
contr <- predict(bst, agaricus.test$data, predcontrib = TRUE)
xgb.plot.shap(agaricus.test$data, contr, model = bst, top_n = 12, n_col = 3)
xgb.ggplot.shap.summary(agaricus.test$data, contr, model = bst, top_n = 12)  # Summary plot

# multiclass example - plots for each class separately:
nclass <- 3
x <- as.matrix(iris[, -5])
set.seed(123)
is.na(x[sample(nrow(x) * 4, 30)]) <- TRUE # introduce some missing values
mbst <- xgboost(data = x, label = as.numeric(iris$Species) - 1, nrounds = nrounds,
                max_depth = 2, eta = 0.3, subsample = .5, nthread = nthread,
                objective = "multi:softprob", num_class = nclass, verbose = 0)
trees0 <- seq(from=0, by=nclass, length.out=nrounds)
col <- rgb(0, 0, 1, 0.5)
xgb.plot.shap(x, model = mbst, trees = trees0, target_class = 0, top_n = 4,
              n_col = 2, col = col, pch = 16, pch_NA = 17)
xgb.plot.shap(x, model = mbst, trees = trees0 + 1, target_class = 1, top_n = 4,
              n_col = 2, col = col, pch = 16, pch_NA = 17)
xgb.plot.shap(x, model = mbst, trees = trees0 + 2, target_class = 2, top_n = 4,
              n_col = 2, col = col, pch = 16, pch_NA = 17)
xgb.ggplot.shap.summary(x, model = mbst, target_class = 0, top_n = 4)  # Summary plot

xgboost documentation built on Sept. 11, 2024, 8:26 p.m.