Quote:
Originally posted by evilpurple
On my todo list is something that also take into account click stats to list galleries with descriptions that are likely to do well higher in my review tool... Was thinking about using bayesian networks or a neural net for it. Will be interesting to see if it's worth the effort 
|
I guess I'm not the only one that's messed around with stuff like this.
I was thinking about Bayes probability for a while but couldn't think of anything that made sense from a script sense. If you get something let me know ;-)
[email protected]
I tried various things - all kinds of crazy ideas - but ultimately the well-known factors like high quality galleries, site design, placement of blinds, loading speed, having a good system to keep trades alive through resets, etc. overwhelmed all the clever stuff I was trying.
Eventually, I ended up with a REALLY simple script that just ordered all the outs according to uniques in. I'm not a great programmer by any means - it's only a hobby - so that decision was because I didn't know how to deal with the cgi hogging all the systems resources as traffic went up. I went from about 3k visitors/day to 80k in 30 days. I had no idea what I was doing programming-wise and slimmed down to the simplest version.