I have three main categories which need to receive a ranking dependent on the ranking provided by user ratings ranging from 1 to 5 (whole numbers) on three qualitative characteristics (with equal importance) assigned to each main categories
I would like to combine the ratings received on each category into a single complex rating, but I also would like to allow users to receive a customized complex ranking on the basis of the relative importance they place on each of the main categories. E.g. if the main categories are Age, Gender, and Height, I would like to come up with percentage weights to assign to each of them in accordance with user perception whether they are very important, important, or somewhat important. For example, if someone considers Age and Gender very important, but height somewhat important, what percentages should I assign? If all three are ranked as very important, important, or somewhat important, do I assign them equal weights of 33.3%? What if we have two categories ranked as somewhat important and one as important? I need a statistical weight distribution for all the possible scenarios.
After a complex rating is delivered, I want to be able to adjust (normalize) it for the following:
Recent ratings should be more relevant than older ones. How to give more weight to more recent reviews?
Quantity of ratings
The number of reviews should affect the overall complex score, the more reviews, the more reliable the overall score. I am wondering how I can account for this.
E.g. If we have user profiles rated as follows: Novice – Advanced – Expert – All-star, how can I can incorporate for this factor too?
I am aiming to have a single equation to arrive at a normalized complex score ranging from 1 to 5. Any suggestions for an algorithm/equation are welcome. I understand that the score still depends on users’ relative perceptions of importance and a set of per-defined criteria, but I want to incorporate them in the most statistically correct way. Thanks to all for your kindness and assistance.