- edited description
Error in Error in ncomp_opt[measure, ijk] <- t.test.process(t(mat.error.rate[[measure_i]][[ijk]]))
Issue #113
resolved
Here is the error we get when running the cs_SRBCT.R script (code chunk number 6: cs_SRBCT.Rnw:90-98 (eval = FALSE)), depending on the seed (also observed for a multilevel plsda using perf), using mixOmics 6.3.0.
Would 6.3.1 fix that bug or is that a new bug?
> set.seed(1111)
> perf.srbct.plsda <- perf(srbct.plsda, validation = "Mfold", folds = 5,
+ progressBar = FALSE, nrepeat = 10)
Error in ncomp_opt[measure, ijk] <- t.test.process(t(mat.error.rate[[measure_i]][[ijk]])) :
number of items to replace is not a multiple of replacement length
> set.seed(2222)
> perf.srbct.plsda <- perf(srbct.plsda, validation = "Mfold", folds = 5,
+ progressBar = FALSE, nrepeat = 10)
Error in ncomp_opt[measure, ijk] <- t.test.process(t(mat.error.rate[[measure_i]][[ijk]])) :
number of items to replace is not a multiple of replacement length
> set.seed(2543)
> perf.srbct.plsda <- perf(srbct.plsda, validation = "Mfold", folds = 5,
+ progressBar = FALSE, nrepeat = 10)
Comments (4)
-
reporter -
This has nothing to do with the update in 6.3.1 so far
-
Fixed in 75c955d
FYI: was coming from error rates being constant in component 8-10 (hence the t-test wasn't working properly)
-
- changed status to resolved
- Log in to comment