add factor biplot

[maxheld83]

this is customary for PCA (and factor) analysis in general.

Might also be helpful for inspecting rotations, though it's a bit visual overload #13

[maxheld83]

this was originally part of #94 which is now about a simpler plot with only the people-variables.

maxheld83@d0ecdf8 starts development on this.

[maxheld83]

like so:

Notice that the above plot is based on a weighted factor scores (as per Brown 1980), so it's not clear if/how this can be interpreted (see #163).

[maxheld83]

I'm pretty happy with q.rotplot() for now, putting this on the back burner.
Also still seems a little information overload.

[maxheld83]

just to document this if I / someone takes this up in future: I think most of the work is done for this; I have adapted the existing ggbiplot() to accept a results object from qmethod(), and it seems to work ok (see above).

Caveats that remain:

• not sure how the weighting and other Q specifics work with a biplot -> make biplot meaningful for flagged, weighted analysis #163
• ggbiplot() is based on SVD, and I have no idea about that, so I cannot be quite sure that my adaptation is correct. Someone must look at this PR -> accept qmethod::qmethod as pcobj vqv/ggbiplot#20
• ggbiplot() is not on CRAN, so it's unclear how it could be used meaningfully -> best way to include / wrap ggbiplot in a CRAN package? vqv/ggbiplot#19
• ggbiplot() would have to be somehow wrapped for qmethod(). Guess that's this issue right here.

Bottom line: we're almost there, and no one should start from scratch.
ggbiplot() is a great package.

[maxheld83]

this might also be relevant, if we're looking for other ways: http://stackoverflow.com/questions/6578355/plotting-pca-biplot-with-ggplot2

[maxheld83]

duplicated here maxheld83/pensieve#82