Get probabilities for some classes.
Arguments
- pred
(Prediction)
Prediction object.- cl
(character)
Names of classes. Default is either all classes for multi-class / multilabel problems or the positive class for binary classification.
Value
(data.frame) with numerical columns or a numerical vector if length of cl
is 1.
Order of columns is defined by cl
.
See also
Other predict:
asROCRPrediction()
,
getPredictionResponse()
,
getPredictionTaskDesc()
,
predict.WrappedModel()
,
setPredictThreshold()
,
setPredictType()
Examples
task = makeClassifTask(data = iris, target = "Species")
lrn = makeLearner("classif.lda", predict.type = "prob")
mod = train(lrn, task)
#> Error in x[0, , drop = FALSE]: incorrect number of dimensions
# predict probabilities
pred = predict(mod, newdata = iris)
#> Error in predict(mod, newdata = iris): object 'mod' not found
# Get probabilities for all classes
head(getPredictionProbabilities(pred))
#> Error in checkClass(x, classes, ordered, null.ok): object 'pred' not found
# Get probabilities for a subset of classes
head(getPredictionProbabilities(pred, c("setosa", "virginica")))
#> Error in checkClass(x, classes, ordered, null.ok): object 'pred' not found