lime {lime} | R Documentation |
Create a model explanation function based on training data
Description
This is the main function of the lime
package. It is a factory function
that returns a new function that can be used to explain the predictions made
by black box models. This is a generic with methods for the different data
types supported by lime.
Usage
## S3 method for class 'data.frame'
lime(
x,
model,
preprocess = NULL,
bin_continuous = TRUE,
n_bins = 4,
quantile_bins = TRUE,
use_density = TRUE,
...
)
## S3 method for class 'character'
lime(
x,
model,
preprocess = NULL,
tokenization = default_tokenize,
keep_word_position = FALSE,
...
)
## S3 method for class 'imagefile'
lime(x, model, preprocess = NULL, ...)
lime(x, model, ...)
Arguments
x |
The training data used for training the model that should be explained. |
model |
The model whose output should be explained |
preprocess |
Function to transform a |
bin_continuous |
Should continuous variables be binned when making the explanation |
n_bins |
The number of bins for continuous variables if |
quantile_bins |
Should the bins be based on |
use_density |
If |
... |
Arguments passed on to methods |
tokenization |
function used to tokenize text for the permutations. |
keep_word_position |
set to |
Value
Return an explainer which can be used together with explain()
to
explain model predictions.
Examples
# Explaining a model based on tabular data
library(MASS)
iris_test <- iris[1, 1:4]
iris_train <- iris[-1, 1:4]
iris_lab <- iris[[5]][-1]
# Create linear discriminant model on iris data
model <- lda(iris_train, iris_lab)
# Create explanation object
explanation <- lime(iris_train, model)
# This can now be used together with the explain method
explain(iris_test, explanation, n_labels = 1, n_features = 2)
## Not run:
# Explaining a model based on text data
# Purpose is to classify sentences from scientific publications
# and find those where the team writes about their own work
# (category OWNX in the provided dataset).
library(text2vec)
library(xgboost)
data(train_sentences)
data(test_sentences)
get_matrix <- function(text) {
it <- itoken(text, progressbar = FALSE)
create_dtm(it, vectorizer = hash_vectorizer())
}
dtm_train = get_matrix(train_sentences$text)
xgb_model <- xgb.train(list(max_depth = 7, eta = 0.1, objective = "binary:logistic",
eval_metric = "error", nthread = 1),
xgb.DMatrix(dtm_train, label = train_sentences$class.text == "OWNX"),
nrounds = 50)
sentences <- head(test_sentences[test_sentences$class.text == "OWNX", "text"], 1)
explainer <- lime(train_sentences$text, xgb_model, get_matrix)
explanations <- explain(sentences, explainer, n_labels = 1, n_features = 2)
# We can see that many explanations are based
# on the presence of the word `we` in the sentences
# which makes sense regarding the task.
print(explanations)
## End(Not run)
## Not run:
library(keras)
library(abind)
# get some image
img_path <- system.file('extdata', 'produce.png', package = 'lime')
# load a predefined image classifier
model <- application_vgg16(
weights = "imagenet",
include_top = TRUE
)
# create a function that prepare images for the model
img_preprocess <- function(x) {
arrays <- lapply(x, function(path) {
img <- image_load(path, target_size = c(224,224))
x <- image_to_array(img)
x <- array_reshape(x, c(1, dim(x)))
x <- imagenet_preprocess_input(x)
})
do.call(abind, c(arrays, list(along = 1)))
}
# Create an explainer (lime recognise the path as an image)
explainer <- lime(img_path, as_classifier(model, unlist(labels)), img_preprocess)
# Explain the model (can take a long time depending on your system)
explanation <- explain(img_path, explainer, n_labels = 2, n_features = 10, n_superpixels = 70)
## End(Not run)