fda.usc (version 0.9.4)

fregre.pls.cv: Functional PLS regression with scalar response using selection of number of PLS components

Description

Functional Regression with scalar response using selection of number of partial leas squares (PLS) comoponents through cross-validation. The algoritm selects the PLS components with best estimates the response. The select is performed by cross-validation (CV) or Model Selecttion Criteria (MSC). After is computing functional regression using the best selection of PLS components.

Usage

fregre.pls.cv(fdataobj, y, kmax=8, criteria = "SIC",...)

Arguments

fdataobj

fdata class object.

kmax

The number of PLS components to include in the model.

y

Scalar response with length n.

criteria

Type of cross-validation (CV) or Model Selecttion Criteria (MSC) applied. Possible values are "CV", "AIC", "AICc", "SIC".

...

Further arguments passed to fregre.pls

Value

• Return:
• pls.opt
  Index of PLS components selected.
• MSC.min
  Minimum Model Selecttion Criteria (MSC) value for the code{pls.opt} components.
• MSC
  Minimum Model Selecttion Criteria (MSC) value for kmax components.
• fregre.pls
  Fitted regression object by the best pls.opt components.

Details

The partial least squares are calculated by plsr function. The method selects the PLS components with minimum MSC criteria by stepwise regression.

References

Preda C. and Saporta G. PLS regression on a stochastic process. Comput. Statist. Data Anal. 48 (2005): 149{-}158.

See Also

See Also as: fregre.pls, summary.fregre.fd and predict.fregre.fd. Alternative method: fregre.pc, fregre.basis and fregre.np.

Examples

data(tecator)
x<-tecator$absorp.fdata
y<-tecator$y$Fat
res1=pls.fdata(x,y,5)
norm.fdata(res1$rotation)
res2=fregre.pls(x,y,c(1:8))
summary(res2)