About this topic

  • Posted by amy_community 2 years ago. There are 8 posts. The latest reply is from byron.
  • This topic is not resolved
  1. Fantastic plugin! It's been great for our website where people are ranking their favourite websites based on the images uploaded - the option chosen to do this is "Sort Images by Rating (Descending High > Low)

    Here's the page:
    http://ppmcommunity.com/ppm-community/

    There seems to be an error when you get half way down the page; at the top of the page are the highest ranked images which is right but then it reverts to images of 1 vote and then towards the end are images with ratings of 2 or more.

    Would greatly appreciate your help!

  2. HI Amy,

    Thanks! Glad you're liking photosmash!!

    I'll need to look into that weird behavior. I'm wondering if somehow the Count of votes is off and is playing into the ranking somehow. It'll have to be fixed in the next release.

    BB

  3. Using the sidebar widget, the rankings (highest to lowest rated) are as so:

    1-4.2 with 6 votes
    2-4.3 with 4 votes
    3-3.9 with 8 votes
    4-3.8 with 6 votes
    5-3.8 with 5 votes
    6-3.3 with 3 votes
    7-3.3 with 6 votes
    8-3.3 with 8 votes

    So - Just numbers 1 and 2 seem to be out.

    I'm only being picky because a visitor will be picky to me, making me explain myself!

    Thanks for the outstanding plugin!

    J

  4. Jess,

    You're very welcome!

    That is actually the preferred behavior for the type of ranking that is being performed. It uses the Bayesian probability formula (http://en.wikipedia.org/wiki/Bayesian_probability) to take into account the number of votes in the ranking. So, in simplest terms, you would want an image that had 4.9 with 1,000 votes to be #1 over an image that had 5.0 with only 1 vote. A straight average would say the 5.0 should win, but that's not usually the preferred method.

    I don't profess to understand the math, but that's what they say :-) And I do see how this makes explaining a little difficult.

    BB

  5. I understand this. It's a very fair system, and I'm happy you implemented it

    But...

    With this theory, 4&5 are inverted - or 6,7&8

    ;)

    Ha! Get out of that one, Mr. Bayesian!

  6. I was hoping you wouldn't notice that! ;-)

    Perhaps there are decimals that are not being shown...that's probably not it, but it sounded good!

    Cheers,
    Byron

  7. Alright. How about this...

    Order the votes from highest average to lowest average, then order this new list into most votes to least votes? It's a decent compramise, I think...

    J

  8. Good suggestion...I'll see if it's something I can control without adding a lot of additional processing to it.

    BB

RSS feed for this topic