error in performance(pred,"tpr","for")

[Xdai8923]

Dear all,

I am running ROCR to get my lncRNA coding potential cutoff by using the provided r script (https://sourceforge.net/projects/rna-cpat/files/Figure3_data/). I have checked my ROCR data, which does have any NA:

is.na(ROCR_data)
predictions Labels
FALSE FALSE
However, in my dataset to start with, there are 4562 labeled with "0" and 37988 labeled with "1". I have shuffled my data by row.
When I run:
perf <- performance(pred,"tpr","for")
It showed the following error:
perf <- performance(pred,"tpr","fpr")
Error in stats::approxfun(x.values.1, y.values.1, method = "constant", :
zero non-NA points

Could you please advise what caused this problem and how should I fix it?

Thank you.
Xiaoxia

[FelixErnst]

@Xdai8923

Hi Xiaoxia,

sorry for the very delayed reply. Without example data, I cannot follow you example and make only a guess. The error message means "only NA values are available". I suspect there is a certain structure in you data causing this.
If you want, please post your data. Otherwise there is little we can do.

Best