https://bz.apache.org/bugzilla/show_bug.cgi?id=62836

--- Comment #3 from Matthias Becht <[email protected]> ---
I'm still not entirely sure about all of Excel's behavior:

- So if all x and y values are the same, Excel just returns that value, no
matter what?

- If the y's are different but all x's are the same, Commons Math works but
gives a different answer than Excel. For instance the test "All x values the
same" (B65; currently skipped) gives -32 but Excel gives 12, which is simply
the average of the two y values (20 and 4). Is this just mathematically
illegal, and Excel chooses to simply average the y values or what is happening
here?

- When there is only one y value, Excel generally just returns that, but it
does still calculate the linear regression if the constant is set to false in
order to force the line to go through zero. I'm assuming it just adds (0,0) as
another point in that case. When there is still only one y value but multiple x
variables to go with it, the y value is always returned, even if the constant
is set to false, which also makes sense since it would need more than just the
point (0,0) to add another sample. I guess the easiest thing to do then would
be to automatically add the point (0,0) if there is only one y and one x value
and the constant is set to false. When there are multiple x variables, the
new_x's, if given, need to be exactly the same dimensions as x (which has to be
either a single column or row), or #REF! is given.

I must admit that I don't know much about the underlying mathematics of this,
so there are probably many obvious things I'm missing.

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to