Data too High Dimension for clustering about eternabrain HOT 3 CLOSED

Nice, excellent.

from eternabrain.

Hi Rohan,

Great work - so you went with the "add NULLs everywhere" approach? Just the PCA alone would be interesting. I'll be following Github to see how it's going. : )

I've mentioned this before, but if you get stuck, taking some smaller question might help make sense of the data at first. For example, look at the first 20 moves instead of the complete set, see if that tells you anything interesting. You could just look at base locations (not at base changes) and look at patterns of initial attention.

Good luck!

Best,
Ben

from eternabrain.

Hi Ben,
Actually, I found that sklearn algorithms don't accept None/NULL as an input, so I'm just using 0's instead. I am also just clustering 5 players' moves, so I can familiarize myself with the data, and get a better sense of the clustering algorithms.

from eternabrain.