sachsmc / plotROC

Interactive ROC plots with ggplot + d3.js.
http://sachsmc.github.io/plotROC
Other
86 stars 15 forks source link

plot mean and standard deviation of ROC curves #78

Closed wbvguo closed 2 years ago

wbvguo commented 2 years ago

Dear plotROC maintainer,

Thanks for maintaining this tool! The package works great in ploting a single ROC curve. I have a small question about multiple ROC curves.

Say I have 100 test datasets (they are repetitive experiments), each of them can generate a ROC curve, I was wondering if I can use plotROC/geom_roc to aggregate them into one ROC plot, with the curve representing the mean and shadowing (or range bar) representing the confidence interval/standard deviation? Or do you have any comments on how to make this plot in ggplot2?

Thanks,

sachsmc commented 2 years ago

Hi Wenbin,

No that is not something that plotROC can do. If you aggregate yourself and have the "mean" ROC curve calculated as a data frame with one column for the FPF and one column for the TPF you can plot it using geom_roc with stat = "identity"

See this section of the vignette for an example of plotting a precalculated ROC:

https://sachsmc.github.io/plotROC/index.html#other-estimation-methods

For error bars or ranges I would probably use geom_ribbon()

Hope that helps.

Best wishes, Michael