predict.aklimate: predict.aklimate

In VladoUzunangelov/aklimate: AKLIMATE : Algorithm for Kernel Learning with Approximating Tree Ensembles

Description

Compute predictions from an aklimate model

Usage

## S3 method for class 'aklimate'
predict(akl_obj, dat, fsets, kernels = NULL,
  store_kernels = FALSE)

Arguments

akl_obj	an AKLIMATE model
dat	samples x features data frame where columns might be of different type
fsets	list of prior knowledge feature sets
kernels	Pre-computed kernels for the MKL prediction part. By default the value is NULL, i.e. test kernels are computed on the fly from the RF models stored in the aklimate object.
store_kernels	Boolean variable indicating whether test_kernels should be stored in output. Default is FALSE.

Value

List with the following elements:

preds

Predictions.
If akl_obj$akl_pars$type is "response", this is a vector class predictions or regression estimates.
If akl_obj$akl_pars$type is "probability" (classification tasks only), it is a matrix with predicted probabilities of membership for each class.

kernels

If store_kernels is TRUE, the test kernels. Default is FALSE.

VladoUzunangelov/aklimate documentation built on Aug. 17, 2020, 4:40 a.m.