statistics functions and NA

[CdeMills]

Hello,
I have data which are segmented. For each segment I compute some model, then
I collate everything. Except that for some segments the result is dubious,
and the associated values for this segment are replaced by NA.

The sum of the components of a vector containing one NA is NA ... so
statistic functions like mean, std, var, all give NA as result.

R as a specific parameter in this case, na.rm which,, when true, means to
skip over NA values. What do you think about adding such behaviour to mean,
std, val and similar ? The idea is to merely keep NA values as in indicator
of problem in the data, and yet extract stats without adding a supplemental
step of filtering away those NA.

Regards

Pascal

--
View this message in context: 
http://octave.1599824.n4.nabble.com/statistics-functions-and-NA-tp4489788p4489788.html
Sent from the Octave - General mailing list archive at Nabble.com.