rchemo (version 0.1-3)

gridcv: Cross-validation

Description

Functions for cross-validating predictive models.

The functions return "scores" (average error rates) of predictions for a given model and a grid of parameter values, calculated from a cross-validation process.

- gridcv: Can be used for any model.

- gridcvlv: Specific to models using regularization by latent variables (LVs) (e.g. PLSR). Much faster than gridcv.

- gridcvlb: Specific to models using ridge regularization (e.g. RR). Much faster than gridcv.

Usage

gridcv(X, Y, segm, score, fun, pars, verb = TRUE)
gridcvlv(X, Y, segm, score, fun, nlv, pars = NULL, verb = TRUE)
gridcvlb(X, Y, segm, score, fun, lb, pars = NULL, verb = TRUE)

Value

Dataframes with the prediction scores for the grid.

Arguments

X

Training X-data (\(n, p\)), or list of training X-data.

Y

Training Y-data (\(n, q\)).

segm

CV segments, typically output of segmkf or segmts.

score

A function calculating a prediction score (e.g. msep).

fun

A function corresponding to the predictive model.

nlv

For gridcvlv. A vector of numbers of LVs.

lb

For gridcvlb. A vector of ridge regulariation parameters.

pars

A list of named vectors. Each vector must correspond to an argument of the model function and gives the parameter values to consider for this argument. (see details)

verb

Logical. If TRUE, fitting information are printed.

Details

Argument pars (the grid) must be a list of named vectors, each vector corresponding to an argument of the model function and giving the parameter values to consider for this argument. This list can eventually be built with function mpars, which returns all the combinations of the input parameters, see the examples.

For gridcvlv, pars must not contain nlv (nb. LVs), and for gridcvlb, lb (regularization parameter \(lambda\)).

Examples

## EXAMPLE WITH PLSR

n <- 50 ; p <- 8
X <- matrix(rnorm(n * p), ncol = p)
y <- rnorm(n)
Y <- cbind(y, 10 * rnorm(n))

K = 3
segm <- segmkf(n = n, K = K, nrep = 1)
segm

nlv <- 5
pars <- mpars(nlv = 1:nlv)
pars
gridcv(
    X, Y, segm,
    score = msep, 
    fun = plskern, 
    pars = pars, verb = TRUE)

gridcvlv(
    X, Y, segm, 
    score = msep, 
    fun = plskern, 
    nlv = 0:nlv, verb = TRUE)
    
## EXAMPLE WITH PLSLDA

n <- 50 ; p <- 8
X <- matrix(rnorm(n * p), ncol = p, byrow = TRUE)
y <- sample(c(1, 4, 10), size = n, replace = TRUE)

K = 3
segm <- segmkf(n = n, K = K, nrep = 1)
segm

nlv <- 5
pars <- mpars(nlv = 1:nlv, prior = c("unif", "prop"))
pars
gridcv(
    X, y, segm, 
    score = err, 
    fun = plslda,
    pars = pars, verb = TRUE)

pars <- mpars(prior = c("unif", "prop"))
pars
gridcvlv(
    X, y, segm, 
    score = err, 
    fun = plslda,
    nlv = 1:nlv, pars = pars, verb = TRUE)

## EXAMPLE WITH RR

n <- 50 ; p <- 8
X <- matrix(rnorm(n * p), ncol = p)
y <- rnorm(n)
Y <- cbind(y, 10 * rnorm(n))

K = 3
segm <- segmkf(n = n, K = K, nrep = 1)
segm

lb <- c(.1, 1)
pars <- mpars(lb = lb)
pars
gridcv(
    X, Y, segm, 
    score = msep, 
    fun = rr, 
    pars = pars, verb = TRUE)

gridcvlb(
    X, Y, segm, 
    score = msep, 
    fun = rr, 
    lb = lb, verb = TRUE)

## EXAMPLE WITH KRR

n <- 50 ; p <- 8
X <- matrix(rnorm(n * p), ncol = p)
y <- rnorm(n)
Y <- cbind(y, 10 * rnorm(n))

K = 3
segm <- segmkf(n = n, K = K, nrep = 1)
segm

lb <- c(.1, 1)
gamma <- 10^(-1:1)
pars <- mpars(lb = lb, gamma = gamma)
pars
gridcv(
    X, Y, segm, 
    score = msep, 
    fun = krr, 
    pars = pars, verb = TRUE)

pars <- mpars(gamma = gamma)
gridcvlb(
    X, Y, segm, 
    score = msep, 
    fun = krr, 
    lb = lb, pars = pars, verb = TRUE)

## EXAMPLE WITH LWPLSR

n <- 50 ; p <- 8
X <- matrix(rnorm(n * p), ncol = p)
y <- rnorm(n)
Y <- cbind(y, 10 * rnorm(n))

K = 3
segm <- segmkf(n = n, K = K, nrep = 1)
segm

nlvdis <- 5
h <- c(1, Inf)
k <- c(10, 20)
nlv <- 5
pars <- mpars(nlvdis = nlvdis, diss = "mahal",
              h = h, k = k)
pars
res <- gridcvlv(
    X, Y, segm, 
    score = msep, 
    fun = lwplsr, 
    nlv = 0:nlv, pars = pars, verb = TRUE)
res