puSummaryThLoop: puSummaryThLoop

In benmack/oneClass: One-class classification in the absence of test data

Description

An example for a PU summary function which derives the maximum performance metric over a set of thresholds. Here 50 thresholds between the medians of the positive and unlabeled hold out predictions.

Usage

puSummaryThLoop(data, lev = NULL, model = NULL, thresholds = NULL,
  maximize = c("puF", "puF1"), returnAll = FALSE, calcAUC = FALSE)

Arguments

data	a data frame or matrix with columns obs, pred, and pos. The first two are the binary observed, predicted outcomes and the latter one the continous outcome for the positive class.
lev	a character vector of factors levels for the response (default is NULL).
model	a character string for the model name (as taken form the method argument of train) (default is NULL).
thresholds	thresholds for which to calcualte the performance metrics
maximize	Charater (vector). The metrics to be maximized. Must correspond to a name of the vector returned by puSummary.
returnAll	return metrics for all thresholds.

Value

values of performance metrics

Examples

## Not run: 
data(bananas)
tuneGrid <- expand.grid(sigma=c(1:10)*rep(c(.1, 1), each=10), 
                       nu=c(0.01, 0.05, 0.10, 0.15, 0.20, 0.25))
model <- trainOcc(x=bananas$tr[, -1], y=bananas$tr[, 1], 
                 method="ocsvm", tuneGrid=tuneGrid)
model <- update(model, aggregatePredictions=TRUE, 
               metric="puFAP")
model <- update(model, aggregatePredictions=TRUE, 
               puSummaryFunction = puSummaryThLoop, 
               metric="thPuFAP")
pairs(model$results[, c("puF", "puFAP", "thPuFAP")])

mp.puF <- modelPosition(model, modRank=1, by="puF")
mp.puFAP <- modelPosition(model, modRank=1, by="puFAP")
mp.thPuFAP <- modelPosition(model, modRank=1, by="thPuFAP")

featurespace(update(model, modRow=mp.puF$row))
featurespace(update(model, modRow=mp.puFAP$row))
featurespace(update(model, modRow=mp.thPuFAP$row))
pred.puF <- predict(update(model, modRow=mp.puF$row), 
                   bananas$x, returnRaster=FALSE)
pred.puFAP <- predict(update(model, modRow=mp.puFAP$row), 
                     bananas$x, returnRaster=FALSE)
pred.thPuFAP <- predict(update(model, modRow=mp.thPuFAP$row), 
                       bananas$x, returnRaster=FALSE)
hist(update(model, modRow=mp.puF$row), pred.puF)
hist(update(model, modRow=mp.puFAP$row), pred.puFAP)
hist(update(model, modRow=mp.thPuFAP$row), pred.thPuFAP)
abline(v=c(0, model$results[mp.thPuFAP$row, "thAP"] ))

## End(Not run)

benmack/oneClass documentation built on Dec. 15, 2020, 7:38 p.m.