Switch to English language

Mainokset

Jos et halua mainoksia sivuille Liity jäseneksi!

All Forums Suggestion and Comments

Rating system


Admin Joined on 14 helmi 2001 Total posts: 1 974	Rating system 23. tammikuuta 2005 19:16 We have changed some parameters in the routine calculating rating - please let us know if you think it is behaving better. Reply Quote
robin Joined on 12 elo 2003 Total posts: 84	Rating system 25. tammikuuta 2005 2:24 Looks much better, yes. Certainly for the couples who compete frequently it makes a lot of sense now... cheers Robin Reply Quote
peter Joined on 13 tammi 2005 Total posts: 468	Rating system 25. tammikuuta 2005 3:25 It is much better now. Certainly now it IS possible that couples dropped out in the early rounds but beating those they were not before, to gain points and not lose them. I still have a feeling that the coefficients are a little too rewarding for couples who win but for whom it is form results. See the City of Birmingham ballroom results - is is completely form result by far, this should not affect their rating, yet all of their rating is modified by a few points - even if by far the result is as expected. Snow Ball standard looks much better, form winners stay as they were. I don'th know if this can be tuned any better than it is now, its getting close to being as good as it is possible. Reply Quote
robbytobby Joined on 19 kesä 2005 Total posts: 6	Re: Rating system 5. helmikuuta 2006 22:50 I have a question regarding the time between a result is added to the database and the time we get a rating for the competition displayed. Sometimes we have to wait about 10 days until the rating is calculated, sometimes we have the rating the next day. is there a reason for this? do you have certain days where the rating is calculated or will the calculation start after a certain number of results has been added? Or are the calculations done accidently? Reply Quote
peter Joined on 13 tammi 2005 Total posts: 468	Re: Rating system 6. helmikuuta 2006 14:03 At the moment the calculation is run manually. The reason for this is that this is very resource-intensive and takes quite a long time: due to the fact that sometimes we add also results from the past means we need to recalculate all the results throughout time - obviously although calculating a single event is quick, we have a lot of events. In the future, we are planning to better optimise this part and implement near-real time calculation (through architecture redesign) for new results. Reply Quote