clarify McNemar

[larskotthoff]

Follow up to #6 -- @berndbischl please have a look.

[berndbischl]

this still is wrong?

\item can be applied to compare the performance of two \textbf{models}
    when the considered performance measure is based on an outer loss with a
    nominal or binary output, e.g.\ accuracy is based on a binary outer loss

the test doesn't refer to any specific measure? you could argue it tests for accuracy, but that's not "e.g." then

[berndbischl]

you also dont test that B=C, as these are absolute values, but that the cell proportions are the same?

[larskotthoff]

Not sure if I get your point. How would you rephrase?

[mlindauer]

@berndbischl could you please look into this again s.t. we can merge this soon. Lars is blocked to record again this video without having the fixed slides.

[mlindauer]

@Ludwig-max @jakob-r can you two somehow comment on the problem on the slides or even fix these?

[jakob-r]

I corrected the part @berndbischl mentioned and double-checked the surroundings. From my side it can be merged.

[mlindauer]

@larskotthoff If you are also happy with the changes, could you please record the videos (in particular the few seconds in the beginning) as soon as possible? We have to deliver the AutoML course soonish, best would be before the new semester starts mid of April. Please take into consideration that also uploading the new videos takes a few days (because the ki-campus has to do it).

[larskotthoff]

Thanks, I've fixed a few typos and requested @berndbischl have a look at it.