iml (version 0.3.0)

PartialDependence: Partial Dependence Plot

Description

PartialDependence computes and plots partial dependence functions of prediction models.

Format

R6Class object.

Usage

pdp = PartialDependence$new(predictor, feature, grid.size = 20, run = TRUE)

plot(pdp) pdp$results print(pdp) pdp$set.feature(2)

Arguments

For PartialDependence$new():

predictor:

(Predictor) The object (created with Predictor$new()) holding the machine learning model and the data.

feature

The feature name or index for which to compute the partial dependencies. Either a single number or vector of two numbers.

grid.size

The size of the grid for evaluating the predictions

run

logical. Should the Interpretation method be run?

Fields

feature.index

The index of the features for which the partial dependence was computed.

feature.name

The names of the features for which the partial dependence was computed.

feature.type

The detected types of the features, either "categorical" or "numerical".

grid.size

The size of the grid.

n.features

The number of features (either 1 or 2)

predictor

The prediction model that was analysed.

results

data.frame with the grid of feature of interest and the predicted \(\hat{y}\). Can be used for creating custom partial dependence plots.

Methods

set.feature

method to get/set feature(s) (by index) fpr which to compute pdp. See examples for usage.

plot()

method to plot the partial dependence function. See plot.PartialDependence

run()

[internal] method to run the interpretability method. Use obj$run(force = TRUE) to force a rerun.

clone()

[internal] method to clone the R6 object.

initialize()

[internal] method to initialize the R6 object.

Details

The partial dependence plot calculates and plots the dependence of f(X) on a single or two features. To learn more about partial dependence plot, read the Interpretable Machine Learning book: https://christophm.github.io/interpretable-ml-book/pdp.html

References

Friedman, J.H. 2001. "Greedy Function Approximation: A Gradient Boosting Machine." Annals of Statistics 29: 1189-1232.

See Also

plot.PartialDependence

Ice for individual conditional expectation plots.

Examples

# NOT RUN {
# We train a random forest on the Boston dataset:
if (require("randomForest")) {
data("Boston", package  = "MASS")
rf = randomForest(medv ~ ., data = Boston, ntree = 50)
mod = Predictor$new(rf, data = Boston)

# Compute the partial dependence for the first feature
pdp.obj = PartialDependence$new(mod, feature = "crim")

# Plot the results directly
plot(pdp.obj)

# Since the result is a ggplot object, you can extend it: 
if (require("ggplot2")) {
 plot(pdp.obj) + theme_bw()
}

# If you want to do your own thing, just extract the data: 
pdp.dat = pdp.obj$results
head(pdp.dat)

# You can reuse the pdp object for other features: 
pdp.obj$set.feature("lstat")
plot(pdp.obj)

# Partial dependence plots support up to two features: 
pdp.obj = PartialDependence$new(mod, feature = c("crim", "lstat"))  
plot(pdp.obj)

# Partial dependence plots also works with multiclass classification
rf = randomForest(Species ~ ., data = iris, ntree=50)
predict.fun = function(object, newdata) predict(object, newdata, type = "prob")
mod = Predictor$new(rf, data = iris, predict.fun = predict.fun)

# For some models we have to specify additional arguments for the predict function
plot(PartialDependence$new(mod, feature = "Sepal.Length"))

# Partial dependence plots support up to two features: 
pdp.obj = PartialDependence$new(mod, feature = c("Sepal.Length", "Petal.Length"))
pdp.obj$plot()   

# For multiclass classification models, you can choose to only show one class:
mod = Predictor$new(rf, data = iris, predict.fun = predict.fun, class = 1)
plot(PartialDependence$new(mod, feature = "Sepal.Length"))
}
# }