SHAP dependence and interaction plots — shap.plot.dependence • SHAPforxgboost

Creates scatter plots showing the relationship between feature values (x-axis) and SHAP values (y-axis). Can display:

Simple dependence: how feature values affect predictions
Colored by another feature: to explore interactions
Interaction effects: when data_int is provided, shows pairwise SHAP interaction values

Usage

shap.plot.dependence(
  data_long,
  x,
  y = NULL,
  color_feature = NULL,
  data_int = NULL,
  dilute = FALSE,
  smooth = TRUE,
  size0 = NULL,
  add_hist = FALSE,
  add_stat_cor = FALSE,
  alpha = NULL,
  jitter_height = 0,
  jitter_width = 0,
  ...
)

Arguments

data_long: the long format SHAP values from shap.prep
x: which feature to show on x-axis, it will plot the feature value
y: which shap values to show on y-axis, it will plot the SHAP value of that feature. y is default to x, if y is not provided, just plot the SHAP values of x on the y-axis
color_feature: which feature value to use for coloring, color by the feature value. If "auto", will select the feature "c" minimizing the variance of the shap value given x and c, which can be viewed as a heuristic for the strongest interaction.
data_int: the 3-dimention SHAP interaction values array. if data_int is supplied, y-axis will plot the interaction values of y (vs. x). data_int is obtained from either predict.xgb.Booster or shap.prep.interaction
dilute: a number or logical, dafault to TRUE, will plot nrow(data_long)/dilute data. For example, if dilute = 5 will plot 20% of the data. As long as dilute != FALSE, will plot at most half the data
smooth: optional to add a loess smooth line, default to TRUE.
size0: point size, default to 1 if nobs<1000, 0.4 if nobs>1000
add_hist: whether to add histogram using ggMarginal, default to TRUE. But notice the plot after adding histogram is a ggExtraPlot object instead of ggplot2 so cannot add geom to that anymore. Turn the histogram off if you wish to add more ggplot2 geoms
add_stat_cor: add correlation and p-value from ggpubr::stat_cor
alpha: point transparancy, default to 1 if nobs<1000 else 0.6
jitter_height: amount of vertical jitter (see hight in geom_jitter)
jitter_width: amount of horizontal jitter (see width in geom_jitter). Use values close to 0, e.g. 0.02
...: additional parameters passed to geom_jitter

Value

be default a ggplot2 object, based on which you could add more geom layers.

Examples

# Example: SHAP dependence plots

# 1. Simple dependence plot: SHAP values vs feature values
shap.plot.dependence(data_long = shap_long_iris, x="Petal.Length",
                     add_hist = TRUE, add_stat_cor = TRUE)
#> `geom_smooth()` using formula = 'y ~ x'
#> `geom_smooth()` using formula = 'y ~ x'


# 2. Show different SHAP values on y-axis
shap.plot.dependence(data_long = shap_long_iris, x="Petal.Length",
                           y = "Petal.Width")
#> `geom_smooth()` using formula = 'y ~ x'


# 3. Color by another feature's values
shap.plot.dependence(data_long = shap_long_iris, x="Petal.Length",
                           color_feature = "Petal.Width")
#> `geom_smooth()` using formula = 'y ~ x'


# 4. Customize x, y, and color features
shap.plot.dependence(data_long = shap_long_iris, x="Petal.Length",
                           y = "Petal.Width", color_feature = "Petal.Width")
#> `geom_smooth()` using formula = 'y ~ x'

# 5. Additional options: histogram, smooth line, data dilution
shap.plot.dependence(data_long = shap_long_iris, x="Petal.Length",
                     y = "Petal.Width", color_feature = "Petal.Width",
                     add_hist = TRUE, smooth = FALSE, dilute = 3)


# Create multiple plots at once
plot_list <- lapply(names(iris)[2:3], shap.plot.dependence, data_long = shap_long_iris)

# SHAP interaction effect plot
# First, prepare the model and interaction data
X_iris = as.matrix(iris[,1:4])
y_iris = as.numeric(iris[[5]]) - 1
dtrain = xgboost::xgb.DMatrix(data = X_iris, label = y_iris)
params = list(learning_rate = 1, min_split_loss = 0, reg_lambda = 0,
              objective = 'reg:squarederror', nthread = 1)
mod1 = xgboost::xgb.train(params = params, data = dtrain,
                          nrounds = 1, verbose = 0)

# Get interaction SHAP values (two methods):
data_int <- shap.prep.interaction(xgb_model = mod1, X_train = X_iris)
# Or directly:
shap_int <- predict(mod1, X_iris, predinteraction = TRUE)

# Plot interaction effects (y-axis shows interaction values)
shap.plot.dependence(data_long = shap_long_iris,
                           data_int = shap_int_iris,
                           x="Petal.Length",
                           y = "Petal.Width",
                           color_feature = "Petal.Width")
#> `geom_smooth()` using formula = 'y ~ x'