plsda_agg: PLSDA with aggregation of latent variables

In mlesnoff/rchemo: Dimension reduction, Regression and Discrimination for Chemometrics

PLSDA with aggregation of latent variables

Description

Ensemblist approach where the predictions are calculated by "averaging" the predictions of PLSDA models built with different numbers of latent variables (LVs).

For instance, if argument nlv is set to nlv = "5:10", the prediction for a new observation is the most occurent level (vote) over the predictions returned by the models with 5 LVS, 6 LVs, ... 10 LVs.

- plsrda_agg: use plsrda.

- plslda_agg: use plslda.

- plsqda_agg: use plsqda.

Usage

plsrda_agg(X, y, weights = NULL, nlv)

plslda_agg(X, y, weights = NULL, nlv, prior = c("unif", "prop"))

plsqda_agg(X, y, weights = NULL, nlv, prior = c("unif", "prop"))

## S3 method for class 'Plsda_agg'
predict(object, X, ...)  

Arguments

X	For the main functions: Training X-data (n, p). — For auxiliary functions: New X-data (m, p) to consider.
y	Training class membership (n). Note: If y is a factor, it is replaced by a character vector.
weights	Weights (n, 1) to apply to the training observations. Internally, weights are "normalized" to sum to 1. Default to NULL (weights are set to 1 / n).
nlv	A character string such as "5:20" defining the range of the numbers of LVs to consider (here: the models with nb LVS = 5, 6, ..., 20 are averaged). Syntax such as "10" is also allowed (here: correponds to the single model with 10 LVs).
prior	The prior probabilities of the classes. Possible values are "unif" (default; probabilities are set equal for all the classes) or "prop" (probabilities are set equal to the observed proportions of the classes in y).
object	A fitted model, output of a call to the main functions.
...	Optional arguments. Not used.

Value

See the examples.

Examples

n <- 50 ; p <- 8
Xtrain <- matrix(rnorm(n * p), ncol = p)
ytrain <- sample(c(1, 4, 10, 2), size = n, replace = TRUE)
#ytrain <- sample(c("a", "10", "d"), size = n, replace = TRUE)
m <- 5
Xtest <- Xtrain[1:m, ] ; ytest <- ytrain[1:m]

############################# PLSRDA-AGG

nlv <- "2:5"
fm <- plsrda_agg(Xtrain, ytrain, nlv = nlv)
names(fm)
res <- predict(fm, Xtest)
names(res)
## Final predictions (after aggregation)
res$pred
err(res$pred, ytest)
## Intermediate predictions (Per nb. LVs)
res$predlv

## Gridscore and gridcv
## Here, there is no sense to use gridscorelv and gridcvlv 

pars <- mpars(nlv = c("1:3", "2:5"))
## Same as:
## pars <- list(nlv = c("1:3", "2:5"))
pars

res <- gridscore(
    Xtrain, ytrain, Xtest, ytest, 
    score = err, 
    fun = plsrda_agg, 
    pars = pars)
res

segm <- segmkf(n = n, K = 3, nrep = 1)
res <- gridcv(
    Xtrain, ytrain, 
    segm, score = err, 
    fun = plslda_agg, 
    pars = pars,
    verb = TRUE)
res

############################# PLSLDA-AGG

nlv <- "2:5"
fm <- plslda_agg(Xtrain, ytrain, nlv = nlv, prior = "unif")
names(fm)
res <- predict(fm, Xtest)
names(res)
res$pred
err(res$pred, ytest)
res$predlv

pars <- mpars(nlv = c("1:3", "2:5"), prior = c("unif", "prop"))
pars
res <- gridscore(
    Xtrain, ytrain, Xtest, ytest, 
    score = err, 
    fun = plslda_agg, 
    pars = pars)
res

segm <- segmkf(n = n, K = 3, nrep = 1)
res <- gridcv(
    Xtrain, ytrain, 
    segm, score = err, 
    fun = plslda_agg, 
    pars = pars,
    verb = TRUE)
res

mlesnoff/rchemo documentation built on April 15, 2023, 1:25 p.m.