Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Method for ranking multiple items based on a population of rankings?

  1. Jul 29, 2009 #1
    Imagine that I want to rank the top 3 foods of all time. I want to ask ten different people to answer this question. So I get answers such as:

    Mashed Potatoes > Artichokes > Carrots
    Beef Jerky > Mashed Potatoes > French Fries

    ...etc. Is there an established method or algorithm for compiling these rankings together? It's clear that you can't simply ask everyone for their 'one favorite', because if everyone says Beets, the results say nothing about the second best food.

    This sort of thing has a lot of applications in sports statistics, ranking teams, players, etc.
     
  2. jcsd
  3. Aug 6, 2009 #2

    EnumaElish

    User Avatar
    Science Advisor
    Homework Helper

  4. Aug 6, 2009 #3
    Interesting. I guess no ranking system is perfect. However, we could still rank them. For instance:
    -what percentage, chose the item as their first choice, their second choice and then their third choice?
    -another option: assign a given point value to each choice. The point value could be equal for all choices, or the first choice could be given a higher point value.
    -or maybe some kind of relative ranking. Whenever x>y then x goes up an equal amount that y goes down.

    Here might be an interesting value, initially, rank items by one of the first two methods, then use some kind of Monte carlo competition, where you randomly select a given preference, then change the items relatively based on that preference (simmillar to the third method). Or alternatively, randomly select two people, and if they each have the same item on the list but ranked differently then adjust the overall rank based on that.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: Method for ranking multiple items based on a population of rankings?
  1. Probability rank-n (Replies: 6)

  2. Rank Vector Entropy (Replies: 0)

Loading...