On my fiction, its prediction of author's gender usually corresponded to the gender of the protagonist. The two exceptions were each a male protagonist and a female author prediction.
When I put in some samples of my academic writing, the predictions all came back as male, except for one, which was a paper I wrote as an undergrad examining whether Homer's depictions of female characters in the Odyssey were sexist; my thesis was no, they are not, so it's interesting to me that that's the one that correctly identified me as female.
Maybe the academic writing conventions are inherently masculine. There's probably a paper in that somewhere, but I'm not going to write it.