is there a way to do classification?

[amueller]

No description provided.

[jmmcd]

Hi @amueller, the short answer is no. Nothing like that was discussed in the FFX paper.

But it does seem like a natural extension, maybe using a logistic transform on the y-prediction. But it wouldn't be very easy, given FFX architecture. I might attempt that or assign a student to it, but not in the near future...

[jmmcd]

I have just seen on http://trent.st/ffx/:

"It can also be used as a classifier (FFXC), by wrapping the output with a logistic map. This has also been used successfully on thousands of industrial problems."

Maybe this is easier than I thought.