Forum: help

RE: Getting the counts straight (re. NA handling) [ Reply ] By: Marc Lijour on 2011-08-12 18:08	[forum:4840]
Thanks for the good work on this package. Suggestions: 1) either discard the NA values in the table, or print a warning on screen. 2) print a summary of the values used as part of the computation (it would make it obvious that NA values have been used). I can't imaging a case for which NA values should be counted, but I am just a newbie.

RE: Getting the counts straight (re. NA handling) [ Reply ] By: Jonathan Swinton on 2011-08-11 14:38	[forum:4837]
I like bugs that fix themselves like this...but I think Vennerable could be smarter even so. Will have a think about this next time I come back to it, suggestions welcome. Thanks for your interest in Vennerable

RE: Getting the counts straight (re. NA handling) [ Reply ] By: Marc Lijour on 2011-08-04 07:08	[forum:4783]
OK got it. NA have no business in the data frame. For the record: > test2 <- lapply(test, na.omit) > venn <- Venn(test2) > venn A Venn object on 3 sets named GT,DM,PA 000 100 010 110 001 101 011 111 0 1 1 1 1 1 0 1 > plot(venn, show = list(Faces=FALSE))

RE: Getting the counts straight (re. NA handling) [ Reply ] By: Marc Lijour on 2011-08-04 06:42	[forum:4782]
Here is the CSV file (upload does not work): GT,DM,PA a,a,a b,, ,c, ,,d e,,e f,f,

Getting the counts straight (re. NA handling) [ Reply ]
By: Marc Lijour on 2011-08-04 06:40

[forum:4781]

I am new here, but this does not seem right (other than that, the package is amazing by the way). The common intersection of 3 sets (111) is counting one too many items. Apparently this is due to the way it handling NA? I am using this correctly? See the attached CSV for a test.

> test <- read.csv("test.csv", na.strings="")
> test
GT DM PA
1 a a a
2 b <NA> <NA>
3 <NA> c <NA>
4 <NA> <NA> d
5 e <NA> e
6 f f <NA>
> venn <- Venn(test)
> venn
A Venn object on 3 sets named
GT,DM,PA
000 100 010 110 001 101 011 111
0 1 1 1 1 1 0 2
> str(venn)
Formal class 'Venn' [package "Vennerable"] with 2 slots
..@ IndicatorWeight : int [1:8, 1:4] 0 1 0 1 0 1 0 1 0 0 ...
.. ..- attr(*, "dimnames")=List of 2
.. .. ..$ : chr [1:8] "000" "100" "010" "110" ...
.. .. ..$ : chr [1:4] "GT" "DM" "PA" ".Weight"
..@ IntersectionSets:List of 8
.. ..$ 000: NULL
.. ..$ 100: chr "b"
.. ..$ 010: chr "c"
.. ..$ 110: chr "f"
.. ..$ 001: chr "d"
.. ..$ 101: chr "e"
.. ..$ 011: NULL
.. ..$ 111: chr [1:2] "a" NA

Look at the last line: what is NA doing there? It is not even common to all three factors!

Thanks for reading this.