Prediction method for objects of class nmfkc

predict.nmfkc generates predictions from an object of class nmfkc, either using the fitted covariates or a new covariate matrix.

When the model was fitted using a formula (Formula Mode), a newdata data frame can be supplied instead of newA; the covariate matrix is then constructed automatically from the stored formula metadata.

Usage

# S3 method for class 'nmfkc'
predict(object, newA = NULL, newdata = NULL, type = "response", ...)

Arguments

object

An object of class nmfkc, i.e., the return value of nmfkc.

newA

Optional. A new covariate matrix to be used for prediction.

newdata

Optional data frame. Only available when the model was fitted using a formula. Covariate columns are extracted automatically using the stored formula metadata. If both newdata and newA are supplied, newdata takes precedence (with a warning).

type

Type of prediction to return. Options are "response" (fitted values matrix), "prob" (soft-clustering probabilities), or "class" (hard-clustering labels based on row names of X).

...

Further arguments passed to or from other methods.

Value

Depending on type: a numeric matrix ("response" or "prob") or a character vector of class labels ("class").

See also

nmfkc, nmfkc.cv

Examples

# Prediction with newA
Y <- matrix(cars$dist, nrow = 1)
A <- rbind(1, cars$speed)
result <- nmfkc(Y, A, rank = 1)
#> Y(1,50)~X(1,1)C(1,2)A(2,50)=XB(1,50)...
#> 0sec
newA <- rbind(1, c(10, 20, 30))
predict(result, newA = newA)
#>          [,1]     [,2]     [,3]
#> [1,] 29.15278 58.14918 87.14558