If you read ?mad you will find this phrase: "median of the absolute deviations from the median"
Note the first word. I think you're too focused on the last word. Peter Ehlers Nair, Murlidharan T wrote: > > -----Original Message----- > From: Deepayan Sarkar [mailto:[EMAIL PROTECTED] > Sent: Monday, September 17, 2007 5:10 PM > To: Nair, Murlidharan T > Cc: [EMAIL PROTECTED] > Subject: Re: [R] MAD > > On 9/17/07, Nair, Murlidharan T <[EMAIL PROTECTED]> wrote: >> I am calculating the median absolute deviation using mad function, and >> it tends to ignore the parameter constant=1, when I am calculating it >> for x=seq(1:5). Am I missing something here? >> >> x<-seq(1:5) >> mad(x)# gives [1] 1.4826 >> mad(x, constant=1)# gives [1] 1 >> #Here is the long form >> dev.from.median<-abs((x-median(x))) >> dev.from.median # Gives [1] 2 1 0 1 2 >> sum(dev.from.median) # Gives [1] 6 >> sum(dev.from.median)/length(x) # Gives [1] 1.2 >> # The long form does not match the output from the function >> >> # When x<-seq(1:10) they match >> x<-seq(1:10) >> dev.from.median<-abs((x-median(x))) >> sum(dev.from.median)/length(x) # Gives 2.5 >> mad(x, constant=1) # Gives 2.5 >> #The long form matches the output from the function >> >> Did I miss anything here? > > yes; mad := Median (not mean) absolute deviation (from the median, by > default). > > -Deepayan > > Indeed, its median and that what I am calculating in the long form. So, > what is that you found I was doing differently? May be I missed your > point. > Thx../M > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > > ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
