sopls_results: Result functions for SO-PLS models
In multiblock: Multiblock Data Fusion in Statistics and Machine Learning

sopls_results

R Documentation

Result functions for SO-PLS models

Description

Standard result functions for SO-PLS (sopls).

Usage

## S3 method for class 'sopls'
predict(
  object,
  newdata,
  ncomp = object$ncomp,
  type = c("response", "scores"),
  na.action = na.pass,
  ...
)

## S3 method for class 'sopls'
coef(object, ncomp = object$ncomp, intercept = FALSE, ...)

## S3 method for class 'sopls'
print(x, ...)

## S3 method for class 'sopls'
summary(
  object,
  what = c("all", "validation", "training"),
  digits = 4,
  print.gap = 2,
  ...
)

classify(object, ...)

## S3 method for class 'sopls'
classify(object, classes, newdata, ncomp, LQ = "LDA", ...)

## S3 method for class 'sopls'
R2(object, estimate, newdata, ncomp = "all", individual = FALSE, ...)

## S3 method for class 'sopls'
RMSEP(object, estimate, newdata, ncomp = "all", individual = FALSE, ...)

pcp(object, ...)

## S3 method for class 'sopls'
pcp(object, ncomp, ...)

## Default S3 method:
pcp(object, X, ...)

cvanova(pred, ...)

## Default S3 method:
cvanova(pred, true, absRes = TRUE, ...)

## S3 method for class 'sopls'
cvanova(pred, comps, absRes = TRUE, ...)

## S3 method for class 'cvanova'
print(x, ...)

## S3 method for class 'cvanova'
summary(object, ...)

## S3 method for class 'cvanova'
plot(x, ...)

## S3 method for class 'sopls'
residuals(object, ...)

Arguments

`object`	A `sopls` object.
`newdata`	Optional new data with the same types of predictor blocks as the ones used for fitting the object.
`ncomp`	An `integer` vector giving the exact components to apply.
`type`	A `character` for `predict` indicating if responses or scores should be predicted (default = "response", or "scores"), for `summary` indicating which type of explained variance to compute (default = "train", alternative = "CV").
`na.action`	Function determining what to do with missing values in `newdata`.
`...`	Additional arguments. Currently not implemented.
`intercept`	A `logical` indicating if coefficients for the intercept should be included (default = FALSE).
`x`	A `sopls` object.
`what`	A `character` indicating if summary should include all, validation or training.
`digits`	The number of digits used for printing.
`print.gap`	Gap between columns when printing.
`classes`	A `character` vector of class labels.
`LQ`	A `character` indicating if 'max' (maximum score value), 'lda' or 'qda' should be used when classifying.
`estimate`	A `character` indicating if 'train', 'CV' or 'test' results should be displayed.
`individual`	A `logical` indicating if results for individual responses should be displayed.
`X`	A `list` of data blocks.
`pred`	An object holding the CV-predicted values (`sopls`, `matrix` or `list` of vectors)
`true`	A `numeric` of true response values for CVANOVA.
`absRes`	A `logical` indicating if absolute (TRUE) or squared (FALSE) residuals should be computed.
`comps`	An `integer` vector giving the exact components to apply.

Details

The parameter ncomp controls which components to apply/extract, resulting in the sequence of components leading up to the specific choice, i.e. ncomp = c(2,2,1) results in the sequence 1,0,0; 2,0,0; 2,1,0; 2,2,0; 2,2,1. Usage of the functions are shown using generics in the examples below. Prediction, regression coefficients, object printing and summary are available through: predict.sopls, coef.sopls, print.sopls and summary.sopls. Explained variances and RMSEP are available through R2.sopls and RMSEP.sopls. Principal components of predictions are available through pcp.sopls. Finally, there is work in progress on classifcation support through classify.sopls.

Value

Returns depend on method used, e.g. predict.sopls returns predicted responses or scores depending on inputs, coef.sopls return regression coefficients, while print and summary methods return the object invisibly.

References

Jørgensen K, Mevik BH, Næs T. Combining designed experiments with several blocks of spectroscopic data. Chemometr Intell Lab Syst. 2007;88(2): 154–166.

Examples

data(potato)
mod <- sopls(Sensory[,1] ~ ., data = potato[c(1:3,9)], ncomp = 5, subset = 1:20)
testset <- potato[-(1:20),]; testset$Sensory <- testset$Sensory[,1,drop=FALSE]
predict(mod, testset, ncomp=c(2,1,2))
dim(coef(mod, ncomp=c(3,0,1))) # <variables x responses x components>
R2(mod, ncomp = c(4,1,2))
print(mod)
summary(mod)

# PCP from sopls object
modMulti <- sopls(Sensory ~ ., data = potato[c(1:3,9)], ncomp = 5, validation = "CV", segment = 5)
(PCP <- pcp(modMulti, c(2,1,2)))
scoreplot(PCP)

# PCP from matrices
preds <- modMulti$validation$Ypred[,,"2,1,2"]
PCP_default <- pcp(preds, potato[1:3])

# CVANOVA
modCV <- sopls(Sensory[,1] ~ ., data = potato[c(1:3,9)], ncomp = 5, validation = "CV", segment = 5)
summary(cva <- cvanova(modCV, "2,1,2"))
plot(cva)

multiblock documentation built on Nov. 18, 2023, 5:06 p.m.