Bug#743418: Variance in LibreOffice
Dnia 2014-04-02, o godz. 16:17:08
"R.B" <badur@op.pl> napisał(a):
> is 3 but it should be 4, because we have 4 arguments
No, it should not; using 3 is correct.
VAR functions computes set (population) variance based on *sample*.
Therefore, it uses formula with correction (sum of squares is divided
by n-1, not n). You can read more about this topic in Wikipedia's
article [0].
Also, if in doubt, always refer to OpenDocument specification [1]. Calc
formulas are described in Part Two, and formula implemented by VAR
function is given in section 6.18.82. We can clearly see "n-1" there.
If you want to compute set (population) variance (i.e. divide sum pf
squares by n), then use VARP function.
The fact that VAR function computes set variance based on sample is
explicitly stated in function's description and help.
Personally, I believe that documentation (help files) should give
insight about particular implementation, but not teach general
subject. From my point of view, it is fine for documentation to assume
that you know what variance is, what sample is, what set (population)
is and how these terms interact with each other. Therefore, I believe
that there is nothing more that could be done to improve LibreOffice in
that particular subject. If you think otherwise, please send your
opinion to upstream - this is in no way Debian specific.
[0] https://en.wikipedia.org/wiki/Variance#Sample_variance
[1] https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=office
direct link to Part Two PDF:
http://docs.oasis-open.org/office/v1.2/OpenDocument-v1.2-part2.pdf
--
Best regards
Mirosław Zalewski
Reply to: