dear list, i have a problem using the adaboost function from the package boost. running the example of leukemia works out very well, but so it does not on my own data. i always get the following 22 warnings: 1: cannot compute exact p-value with ties in: wilcox.test.default(x[which(y == 0)], x[which(y == 1)]) can anybody please help me with the meaning of this warning. thnak you stefan [[alternative HTML version deleted]]
Try this: options(warn = 2) Now rerun it and it will fail with an error at the point of the warning and you can issue a: traceback() to find the exact point at which it failed so you can investigate further. On 3/29/06, Sven Schaltenbrand <sven at schaltenbrand.de> wrote:> dear list, > > i have a problem using the adaboost function from the package boost. > running the example of leukemia works out very well, but so it does not on > my own data. > > i always get the following 22 warnings: > > 1: cannot compute exact p-value with ties in: wilcox.test.default(x[which(y > == 0)], x[which(y == 1)]) > > can anybody please help me with the meaning of this warning. > > thnak you > > stefan > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html >
Sven Schaltenbrand wrote:> dear list, > > i have a problem using the adaboost function from the package boost. > running the example of leukemia works out very well, but so it does not on > my own data. > > i always get the following 22 warnings: > > 1: cannot compute exact p-value with ties in: wilcox.test.default(x[which(y > == 0)], x[which(y == 1)]) > > can anybody please help me with the meaning of this warning.As far as I understand, adaboost() uses the Wilcoxon test statistic as some measure to do variable preselection. Since you can adjust how many "features" should be selected, it seems to be less important to obtain *exact* p-values and you can ignore the warning. Nevertheless it makes sense to look into your data where the ties come from ... Uwe Ligges> thnak you > > stefan > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html