R devel - Jul 2004 - Accuracy in summary (PR#7121)

Full_Name: Sanford Weisberg
Version: 1.9.1
OS: Win XP
Submission from: (NULL) (160.94.148.2)

> wm <-
read.table(url("http://www.stat.umn.edu/~sandy/wmdata0.txt"),
header=TRUE)
> mean(wm$Spd1)
[1] 7.7773
> summary(wm$Spd1)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  0.222   4.780   7.550   7.780  10.200  21.600 

The mean of this variable DOES NOT ROUND to the value shown in the summary
output. 
> options(digits=6)
> mean(wm$Spd1)
[1] 7.77733
> summary(wm$Spd1)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  0.222   4.780   7.550   7.780  10.200  21.600

This is still wrong.  Also, the minimum value is 0.2221, which should have been
printed in full given digits=6.

On Tue, 20 Jul 2004 15:54:49 +0200 (CEST), sandy@stat.umn.edu wrote :
>Full_Name: Sanford Weisberg
>Version: 1.9.1
>OS: Win XP
>Submission from: (NULL) (160.94.148.2)
>
>
>> wm <-
read.table(url("http://www.stat.umn.edu/~sandy/wmdata0.txt"),
>header=TRUE)
>> mean(wm$Spd1)
>[1] 7.7773
>> summary(wm$Spd1)
>   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
>  0.222   4.780   7.550   7.780  10.200  21.600 
>
>The mean of this variable DOES NOT ROUND to the value shown in the summary
>output. 
This is not a bug. summary.default() is behaving as documented.  You
asked (by default) for 3 significant figures, and that's what you got.
>
>> options(digits=6)
>> mean(wm$Spd1)
>[1] 7.77733
>> summary(wm$Spd1)
>   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
>  0.222   4.780   7.550   7.780  10.200  21.600
>
>This is still wrong.  Also, the minimum value is 0.2221, which should have
been
>printed in full given digits=6.
No, see the man page.

Duncan Murdoch