Fast Algorithms to Bootstrap Receiver Operating Characteristics
Implements a very fast C++ algorithm to quickly bootstrap receiver
operating characteristics (ROC) curves and derived performance metrics,
including the area under the curve (AUC) and the partial area under the curve as well as
the true and false positive rate. The analysis of paired receiver operating curves is supported as well,
so that a comparison of two predictors is possible. You can also plot the
results and calculate confidence intervals. On a typical desktop computer the time needed for
the calculation of 100000 bootstrap replicates given 500 observations requires time on the
order of magnitude of one second.
title: "Readme for fbroc"
author: "Erik Peter"
fbroc is intended for the fast bootstrapping of ROC curves, so that the
package can be used for simulation studies and shiny applications. It allows for the analysis and
comparison of ROC curves. To achieve the necessary performance all critical algorithms are implemented in C++.
On a typical desktop computer the time needed for the calculation of 100000 bootstrap replicates
given 500 observations requires time on the order of magnitude of one second.
To try out the package you can visit the package website.
A shiny interface for this package is hosted there.
- latest released version:
- latest development version:
- install and load package devtools
author: "Erik Peter"
date: "Saturday, 18th June, 2015"
- Partial AUCs over both TPR and FPR ranges can be calculated
- You can now adjust text size for plots
- In the ROC plot the (partial) AUC can now optionally be shown instead of confidence regions
- The location of the text showing the performance in the ROC plot has been shifted downwards and
to the left
- Fixed broken plot function for single ROC curves when showing a metric
- Allows the comparison of two paired ROC curves
- Bad defaults caused plotting to fail with a large number of negative samples
- perf.roc is now deprecated. Use the new S3 generic perf instead
- conf.roc is now deprecated. Use the new S3 generic conf instead
- fixed a off-by-one pointer error
- Allow uncached bootstrap of the ROC curve to avoid memory issues, this now the new default
- New performance metrices: TPR at fixed FPR and FPR at fixed TPR
- Stand-alone function to find thresholds
calculate.thresholds was removed. To calculate thresholds
please call boot.roc and look at list item
roc of the outpot
- Smarter default for the number of steps in
- Smarter default for the number of bins in
- Completely refactored C++ code for improved maintability
boot.tpr.at.fpr now works properly
- For duplicated predictions not all relevant thresholds were found reliably, this was fixed