R: Predict response on new data

predict.graper {graper}

R Documentation

Predict response on new data

Description

Function to predict the response on a new data set using a fitted graper model.

Usage

## S3 method for class 'graper'
predict(object, newX, type = c("inRange", "response",
  "link"), ...)

Arguments

`object`	fitted graper model as obtained from `graper`
`newX`	Predictor matrix of size n_test (number of new test samples) x p (number of predictors) (same feature structure as used in `graper`)
`type`	type of prediction returned, either: response:returns the linear predictions for linear regression and class probabilities for logistic regression link:returns the linear predictions inRange:returns linear predictions for linear and class memberships for logistic regression
`...`	other arguments

Value

A vector with predictions.

Examples

# create data
dat <- makeExampleData()
# split data into train and test sets of equal size
ntrain <- dat$n / 2
# fit the model to the train data
fit <- graper(dat$X[seq_len(ntrain), ],
              dat$y[seq_len(ntrain)], dat$annot)
# make predictions on the test data
ypred <- predict(fit, dat$X[seq_len(ntrain) + dat$n / 2, ])

# create data for logistic regression
dat <- makeExampleData(response="bernoulli")
# split data into train and test sets of equal size
ntrain <- dat$n / 2
# fit the graper model for a logistic model
fit <- graper(dat$X[seq_len(ntrain), ],
              dat$y[seq_len(ntrain)],
              dat$annot, family="binomial")
# make predictions on the test data
ypred <- predict(fit, dat$X[seq_len(ntrain) + dat$n / 2, ], type = "inRange")

[Package graper version 1.4.0 Index]