Eloi Ribeiro wrote: > Hi, > > We'r trying to do a chisq test from Python using Rpy but the p-value > result is different depending if we do it directly in R or Python. > So the mistake must be in the way that data is defined. > Can someone help us in this matter? > Thanks in advance. > > Outputs: > ----------------------------------------------------------------------------- > R output
Confirmed on R 2.2.1 (yes, its old!) > m<-matrix(c(386,113,385,117,383,117),nrow=3,byrow=T) > chisq.test(m)$p.value [1] 0.9535284 But notice: > chisq.test(c(386,385,383), c(113,117,117))$p.value [1] 0.2231302 Warning message: Chi-squared approximation may be incorrect in: chisq.test(c(386, 385, 383), c(113, 117, 117)) > ----------------------------------------------------------------------------- > > Python / Rpy output Confirmed on an older version of RPy: >>> from rpy import * >>> c1 = [386,385,383] >>> c2 = [113,117,117] >>> r.chisq_test(c1,c2)["p.value"] Warning message: Chi-squared approximation may be incorrect in: function (x, y = NULL, correct = TRUE, p = rep(1/length(x), length(x)), 0.22313016014842982 You are not passing the matrix correctly. One option that works is: >>> r.chisq_test(r.cbind([386,385,383],[113,117,117]))["p.value"] 0.9535284154083411 I would suggest building an array in python (using Numeric or numpy) instead. Peter ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ rpy-list mailing list rpy-list@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rpy-list
