search

plesager / ece3-postproc

Suite of processing tools for EC-Earth3 output
5 stars 8 forks source link

cannot compute exact p-value #23

Open plesager opened 6 years ago

plesager commented 6 years ago

I got this Warning message:

In ks.test(index_exp1, index_exp2) : cannot compute exact p-value with ties

How should it be interpreted? Can it be ignored or does the code need to deal with this warning? I did not spot anything spurious in the plots.

plesager commented 6 years ago

More generally I got warning that are not printed:

There were 13 warnings (use warnings() to see them)

should we worry about them? I am not familiar with R, so I am not quite sure what its philosophy is with numerical warnings.

etiennesky commented 6 years ago

warnings are not usually bad, but if you can run the commands in an interactive session try running warnings() just after running the command. Post the output here and we can check.

fmassonn commented 6 years ago

Hey, since it's Martin Ménégoz who coded this script, I will transfer the request to him directly and invite him to answer here.

etiennesky commented 6 years ago

or if you can't run interactively, try adding a warnings() statement just after the ks.test() statement.

mmenegoz commented 6 years ago

I think that we do not have to worry about these warnings. It just means that the computing of the p-value has been done with some simplifications, as explained there:

https://stats.stackexchange.com/questions/113032/ks-test-and-ks-boot-exact-p-values-and-ties

plesager commented 6 years ago

I added

options(warn=1)

in the code to get the warnings printed. I have only two kind of warnings, the first one already mentioned in the OP:

In ks.test(index_exp1, index_exp2) : cannot compute exact p-value with ties

which I understand we should not worry about, and the second one:

Warning: ! Warning: Parameter 'sizetit' is obsolete. Use 'title_scale' instead.

which is fixed in aa3a75b