[Libreoffice-bugs] [Bug 117318] Wrong SIGMA SLOPE value calculated with LINEST function

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=117318

--- Comment #3 from Pablo Wagner <pablofabian.wagnerboian at gmail.com> ---
(In reply to Pablo Wagner from comment #2)
> (In reply to raal from comment #1)
> > Hello, please could you specify which formulas are wrong a what are expected
> > values? I compared values from range B35:C39 (only here I see function
> > linest) with excel and the values are the same. Thank you
> 
> I've already tried in Excel and it's also wrong. 
> 
> The expected value for the variance of sigma_slope should be (J52)^2, not
> B36^2. M52^2 is the variance of sigma_slope calculated step by step,
> assuming the intercept with the Y-axis is zero. As can be seen, the LINEST
> function is returning the statistic sigma_slope as if the intercept with the
> Y-axis is zero, which, in this case, isn't zero (close, but not zero).
> 
> The link in the OP points to the source code where that value is calculated,
> and apparently it's wrong.


The expected value for the variance of the slope should be (J52)^2, not B36^2.
M52^2 is the variance of the slope calculated step by step, assuming the
intercept with the Y-axis is zero. As can be seen, the LINEST function is
returning the statistic sigma_slope as if the intercept with the Y-axis is
zero, which, in this case, isn't zero (close, but not zero).

EDITED: Wrong quantity (variance sigma_slope isn't right)

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice-bugs/attachments/20180428/16a1eab3/attachment-0001.html>