plot_PPPvsTPR: Plot of the probability of positive prediction versus the...
In benmack/oneClass: One-class classification in the absence of test data

Description Usage Arguments Details Value Examples

Assumption that the true positive rate can be estimated well from the positive data, it is reasonable to assume that for a given true positive rate a model which leads to a lower probability of positive predictions is more accurate. This is because it will lead to lower fals positive predictions.

1 2	plot_PPPvsTPR(x, identifyPoints = FALSE, add = FALSE, highlightBest = TRUE, xlim = c(0, 1), ylim = c(0, 1), ...)

`x`	a `trainOcc` object
`identifyPoints`	logical with default `FALSE`. set to `TRUE` and click close to the points in the plot if you want to find out the model rows
`add`	logical, add the plot to another plot
`highlightBest`	logical with default set to `TRUE` which highlights the model ranked highest by the selection criteria, i.e. the one in `x$metric`
`xlim`	x axis limits
`ylim`	y axis limits
`...`	other parameters passed to plot (if `add` is `FALSE`) or points (if `add` is `TRUE`)

plot_PPPvsTPR

a scatterplot. the point corresponding to the final model in x is highlighted.

## Not run: 
data(bananas)
### get some models: 
# one-class svm
ocsvm <- trainOcc (x = bananas$tr[, -1], y = bananas$tr[, 1], method="ocsvm", 
                   tuneGrid=expand.grid(sigma=c(seq(0.01, .09, .02), seq(0.1, .9, .1)), 
                                        nu=seq(.05,.55,.1)) )
# biased svm
biasedsvm <- trainOcc (x = bananas$tr[, -1], y = bananas$tr[, 1], method="bsvm", 
                       tuneGrid=expand.grid(sigma=c(0.1, 1), 
                                            cNeg=2^seq(-4, 2, 2), 
                                            cMultiplier=2^seq(4, 8, 2) ) )
# compare with PPPvsTPR-plot
plot_PPPvsTPR(ocsvm)
plot_PPPvsTPR(biasedsvm, add=TRUE, col="red")

## End(Not run)