Rating bar: Difference between revisions
Jump to navigation
Jump to search
Juha Villman (talk | contribs) (→Scale: added link to discussion) |
(beta distribution for rating) |
||
| (One intermediate revision by the same user not shown) | |||
| Line 2: | Line 2: | ||
Rating bar is an extension which can produce bar and star shaped ratings to a wiki page. More information about extension can be found from http://www.mediawiki.org/wiki/Extension:Rating_Bar | Rating bar is an extension which can produce bar and star shaped ratings to a wiki page. More information about extension can be found from http://www.mediawiki.org/wiki/Extension:Rating_Bar | ||
==Updating ratings with Bayes== | |||
Peer rating is described by a beta probability distribution with parameters a and b. There should be a prior that is rather flat but clearly have a tail to the right. The mean could be e.g. 0.3. Then, there should be an analytical solution to what happens to a and b when one additional vote comes in. The result is shown on the rating bar as the mean but also as the 90 % confidence interval. Mean should be a symbol and the coloured bar should stand for the confidence interval, not mean like now. | |||
{{todo|Katsokaapas tätä ideaa ja kommentoikaa. Kunhan rauhoittuvat työt, voisi tämänkin miettiä. Teemulle luonnostaan lankeaa tuo analyyttinen ratkaisu. Tietotekniseen rating barin viilaaminen voisi kysyä Teemu K:ta, ellei joku muu tunne koodia riittävästi.|Juha Villman, Teemu Rintala, Einari Happonen|project=Opasnet}} | |||
==Scale== | ==Scale== | ||
See discussion about the scale {{disclink|Email from Jouni}} | See discussion about the scale {{disclink|Email from Jouni}} and [[peer rating]]. | ||
==See also== | |||
* [[Peer rating]] | |||
Latest revision as of 03:52, 25 March 2011
| [show] |
|---|
Rating bar is an extension which can produce bar and star shaped ratings to a wiki page. More information about extension can be found from http://www.mediawiki.org/wiki/Extension:Rating_Bar
Updating ratings with Bayes
Peer rating is described by a beta probability distribution with parameters a and b. There should be a prior that is rather flat but clearly have a tail to the right. The mean could be e.g. 0.3. Then, there should be an analytical solution to what happens to a and b when one additional vote comes in. The result is shown on the rating bar as the mean but also as the 90 % confidence interval. Mean should be a symbol and the coloured bar should stand for the confidence interval, not mean like now.
TODO: {{#todo:Katsokaapas tätä ideaa ja kommentoikaa. Kunhan rauhoittuvat työt, voisi tämänkin miettiä. Teemulle luonnostaan lankeaa tuo analyyttinen ratkaisu. Tietotekniseen rating barin viilaaminen voisi kysyä Teemu K:ta, ellei joku muu tunne koodia riittävästi.|Juha Villman, Teemu Rintala, Einari Happonen|Opasnet}}
Scale
See discussion about the scale D↷ and peer rating.