predict_surrogate: Instance Level Surrogate Models

In ModelOriented/DALEXtra: Extension for 'DALEX' Package

Instance Level Surrogate Models

Description

Interface to different implementations of the LIME method. Find information how the LIME method works here: https://ema.drwhy.ai/LIME.html.

Usage

predict_surrogate(explainer, new_observation, ..., type = "localModel")

predict_surrogate_local_model(
  explainer,
  new_observation,
  size = 1000,
  seed = 1313,
  ...
)

predict_model.dalex_explainer(x, newdata, ...)

model_type.dalex_explainer(x, ...)

predict_surrogate_lime(
  explainer,
  new_observation,
  n_features = 4,
  n_permutations = 1000,
  labels = unique(explainer$y)[1],
  ...
)

## S3 method for class 'predict_surrogate_lime'
plot(x, ...)

predict_surrogate_iml(explainer, new_observation, k = 4, ...)

Arguments

explainer	a model to be explained, preprocessed by the 'explain' function
new_observation	a new observation for which predictions need to be explained
...	other parameters that will be passed to
type	which implementation of thee LIME method should be used. Either localModel (default), lime or iml.
size	will be passed to the localModel implementation, by default 1000
seed	seed for random number generator, by default 1313
x	an object to be plotted
newdata	alias for new_observation
n_features	will be passed to the lime implementation, by default 4
n_permutations	will be passed to the lime implementation, by default 1000
labels	will be passed to the lime implementation, by default first value in the y vector
k	will be passed to the iml implementation, by default 4

Value

Depending on the type there are different classess of the resulting object.

References

Explanatory Model Analysis. Explore, Explain and Examine Predictive Models. https://ema.drwhy.ai/

ModelOriented/DALEXtra documentation built on June 28, 2023, 5:01 p.m.