Hi all, we conducted a DEA and wanted to correct our efficiency for bias by bootstrapping with R version 3.0.1. With this version, the package "FEAR" does not work anymore. Therefore, we used "dea.boot" for bootstrapping. This, however, returns us negative bias-corrected efficiencies. In the past, we have used "boot.fear" and this never delivered any negative bias-corrected efficiencies. Is this a bug of dea.boot or of R 3.0.1? How can we solve this? Thanks and best regards, Vera -- View this message in context: http://r.789695.n4.nabble.com/Problem-with-dea-boot-under-R-3-0-1-tp4669964.html Sent from the R help mailing list archive at Nabble.com.
... we used the following syntax: b.x1a.y1<-dea.boot(x1a,y1,NREP=1000,RTS="vrs", ORIENTATION="in") Vera -- View this message in context: http://r.789695.n4.nabble.com/Problem-with-dea-boot-under-R-3-0-1-tp4669964p4670271.html Sent from the R help mailing list archive at Nabble.com.
Dear Vera, I had a similar problem once and as far as I can remember the reason were some negative inputs or outputs. # Check for negative values x1a.neg <- apply(x1a, 1, function(x) any(x<0)) y1.neg <- apply(y1, 1, function(x) any(x<0)) exclude <- x1a.neg | y1.neg # Exclude negative rows x1a.neg <- x1a.neg[!exclude,] y1.neg <- y1.neg[!exclude,] # Try again! Daniel -- View this message in context: http://r.789695.n4.nabble.com/Problem-with-dea-boot-under-R-3-0-1-tp4669964p4673363.html Sent from the R help mailing list archive at Nabble.com.