logitFD.fpc: Filtered Functional Principal Component Logistic Regression...

In logitFD: Functional Principal Components Logistic Regression

Description

Fit of the Filtered Functional Principal Component Logistic Regression model with selected Functional Principal Components included in the model according their explained variability.

Usage

logitFD.fpc(Response, FDobj = list(), ncomp = c(), nonFDvars = NULL)

Arguments

Response	Binary (numeric or character) vector of observations of the response variable.
FDobj	List of functional objects from fda package with the curves of the predictor functional variables.
ncomp	Numeric vector with the number of components to be considered for each functional predictor. The vector has equal lenght than FDobj.
nonFDvars	Matrix or data frame with the observations of non-functional variables.

Value

glm.fit	glm object of the fitted model. The object allows to use the summary() function for printing a summary of the fit, the anova() function to produce an analysis of variance table, and to extract useful features as coefficients, effects, fitted.values or residuals
Intercept	Intercept estimated parameter
betalist	List of functional objects (fdobj) of fda package with the estimated parameter functions. Each element of the list corresponds to the associated functional predictor located in the same position of FDobj. All methods of fd package can be used as the plot() function among others.
PC.variance	List of data frames with explained variability of functional principal components of functional predictors. Each element of the list contains the acumulation variance matrix corresponding to the associated functional variable in the same position.
ROC.curve	Object of the roc() function of the pROC package for prediction ability testing of the model. The object can be printed, plotted, or passed to many other functions

Author(s)

Carmen Lucia Reina <[e-mail]> and Manuel Escabias <[e-mail]>

Examples

library(fda.usc)
data(aemet)
Temp<-aemet$temp$data
Prec<-exp(aemet$logprec$data)
StationsVars<-aemet$df[,c("ind","altitude","longitude","latitude")]
StationsVars$North<-c(1,1,1,1,0,0,0,0,1,1,0,0,0,0,0,1,1,1,0,0,1,0,0,0,0,1,0,0,1,1,1,1,1,
0,0,0,1,1,1,1,1,1,1,1,1,0,0,0,0,1,1,1,1,1,0,0,0,0,0,0,0,0,1,1,1,0,0,1,1,1,1,1,1)
TempMonth<-matrix(0,73,12)
PrecMonth<-matrix(0,73,12)
for (i in 1:nrow(TempMonth)){
  TempMonth[i,1]<-mean(Temp[i,1:31])
  PrecMonth[i,1]<-mean(Prec[i,1:31])
  TempMonth[i,2]<-mean(Temp[i,32:59])
  PrecMonth[i,2]<-mean(Prec[i,32:59])
  TempMonth[i,3]<-mean(Temp[i,60:90])
  PrecMonth[i,3]<-mean(Prec[i,60:90])
  TempMonth[i,4]<-mean(Temp[i,91:120])
  PrecMonth[i,4]<-mean(Prec[i,91:120])
  TempMonth[i,5]<-mean(Temp[i,121:151])
  PrecMonth[i,5]<-mean(Prec[i,121:151])
  TempMonth[i,6]<-mean(Temp[i,152:181])
  PrecMonth[i,6]<-mean(Prec[i,152:181])
  TempMonth[i,7]<-mean(Temp[i,182:212])
  PrecMonth[i,7]<-mean(Prec[i,182:212])
  TempMonth[i,8]<-mean(Temp[i,213:243])
  PrecMonth[i,8]<-mean(Prec[i,213:243])
  TempMonth[i,9]<-mean(Temp[i,244:273])
  PrecMonth[i,9]<-mean(Prec[i,244:273])
  TempMonth[i,10]<-mean(Temp[i,274:304])
  PrecMonth[i,10]<-mean(Prec[i,274:304])
  TempMonth[i,11]<-mean(Temp[i,305:334])
  PrecMonth[i,11]<-mean(Prec[i,305:334])
  TempMonth[i,12]<-mean(Temp[i,335:365])
  PrecMonth[i,12]<-mean(Prec[i,335:365])
  }
FourierBasis<-create.fourier.basis(rangeval = c(1,12),nbasis=7)
BsplineBasis<-create.bspline.basis(rangeval = c(1,12),nbasis=8)
TempMonth.fd<-Data2fd(argvals = c(1:12), y=t(TempMonth),basisobj = FourierBasis)
PrecMonth.fd<-Data2fd(argvals = c(1:12), y=t(PrecMonth),basisobj = BsplineBasis)
logitFD.fpc(Response=StationsVars$North,FDobj=list(TempMonth.fd,PrecMonth.fd),
 ncomp = c(3,4),nonFDvars = StationsVars[,c("altitude","longitude")])

logitFD documentation built on Jan. 10, 2022, 9:06 a.m.