?

Log in

No account? Create an account
Queue
August 11th, 2002
05:00 pm

[Link]

Previous Entry Share Flag Next Entry
Jotto stats, part 6
So I talked to a friend who knows all about statistics and such. With some off-the-cuff calculations, he says that my results are probably statistically significant, which means that the dumb AI is better than the "smart" AI. At some point, I may play around some more with Jotto AIs, but I think I'm done with it for now. Time for the next project, I think.

Tags:

(2 comments | Leave a comment)

Comments
 
[User Picture]
From:mattlistener
Date:August 12th, 2002 09:47 am (UTC)
(Link)
That's odd! What was your programming approach?
[User Picture]
From:queue
Date:August 12th, 2002 04:01 pm (UTC)
(Link)
I used your idea that you emailed me about back in January. So, for each word in the dictionary, score it against every other word. Keep track of how many 0s, 1s, 2s, 3s, 4s, and 5s you get. Here is where I went off on my own. You said that you wanted to pick a word that had the most even distribution of scores, so that whatever score you got would need the fewest guesses to narrow it down. I decided to implement this by taking the average "height" of the scores (the average of the numbers of each score). For any of those that had the lowest average, take the standard deviation, and then pick one of the ones that has the lowest standard deviation.

I'm not sure why this didn't work. If you'd like to see the code, let me know, and I'd be happy to share so that we can try to get some insight into what's going on.
My Website Powered by LiveJournal.com