Jump to content
  • 0

weighted favorites score for caches


jasper

Question

First of all: thanks for the great app! Highly appreciated. 

When I search for caches, I sometimes want to see "nice" caches. So with a lot of favorite points. When I order or filter, I am not completely satisfied by the result. A cache that is viewed by thousands of people and got 30 favo's so far is weighted higher then a new cache that is seen by 25 (premium) people and all of them gave it a favorite point. A solution might be to order by the percentage favorites so the percentage of people that gave the cache a favo. The problem with that approach is that a cache that is only found once and got a favo, will have a score of 100% and will be on the top of the list.

In my opinion, a weighted rating for both filtering and sorting would be ideal. A good formulae could be:

weighted rating = ( cache_favo_percentage * nr_premium_finds + avg_favo_percentage * 30 ) / ( nr_premium_finds + 30 )

or simplified:

weighted rating = ( nr_favos + 3 ) / ( nr_premium_finds + 30 )

(as avg_favo_percentage doesn't have to be very accurate and we know that the maximum of this number is 10%)

The idea is that with no finds, the weighting rating is the average favo percentage of all the caches in the world. This is a good starting point as we don't have any other information for this cache. The more finds this cache gets, the more accurate we can determine the actual favo percentage, so the global average will be weighted less and less. IMDB.com uses something similar for scoring their movies. See the attachment.

Note that this formula might seem complex, but it is trivial to compute. And I think it is relatively easy for users to understand. Maybe you can call it the cachly weighted favo score in the filter and sort screens. (or the jasper favo score if you like that better ;-) )

Here are some examples on the differences. The third column shows the favo percentage. The last column is the score based on the weighted percentage. 

finds favos percentage
weighted percentage
0 0 NaN 10%
1 0 0% 10%
1 1 100% 13%
3 0 0% 9%
3 1 33% 12%
3 3 100% 18%
10 1 10% 10%
10 3 30% 15%
10 5 50% 20%
100 0 0% 2%
100 3 3% 5%
100 20 20% 18%
1000 0 0% 0%
1000 8 1% 1%
1000 20 2% 2%

Ordered by favos (which is common now):

finds favos percentage
weighted percentage
100 20 20% 18%
1000 20 2% 2%
1000 8 1% 1%
10 5 50% 20%
3 3 100% 18%
10 3 30% 15%
100 3 3% 5%
1 1 100% 13%
3 1 33% 12%
10 1 10% 10%
0 0 NaN 10%
1 0 0% 10%
3 0 0% 9%
100 0 0% 2%
1000 0 0% 0%

Ordered by favo percentage (This is already better as is gives less eight on the cache with 1000 finds and 8 favos.):

finds favos percentage
weighted percentage
0 0 NaN 10%
3 3 100% 18%
1 1 100% 13%
10 5 50% 20%
3 1 33% 12%
10 3 30% 15%
100 20 20% 18%
10 1 10% 10%
100 3 3% 5%
1000 20 2% 2%
1000 8 1% 1%
1 0 0% 10%
3 0 0% 9%
100 0 0% 2%
1000 0 0% 0%

Ordered by cachly weighted favo score:

finds favos percentage
weighted percentage
10 5 50% 20%
3 3 100% 18%
100 20 20% 18%
10 3 30% 15%
1 1 100% 13%
3 1 33% 12%
0 0 NaN 10%
10 1 10% 10%
1 0 0% 10%
3 0 0% 9%
100 3 3% 5%
100 0 0% 2%
1000 20 2% 2%
1000 8 1% 1%
1000 0 0% 0%

Screen Shot 2016-04-09 at 22.58.55 .png

Link to comment
Share on other sites

6 answers to this question

Recommended Posts

  • 0

The problem with this is that it wouldn't be possible to do this with Live searching, since this uses the API and there is no API request or parameters that can do what you are asking. 

It could be done in Offline, but at the moment isn't something I have planned.

3 hours ago, rragan said:

Project-GC has this. It is called the Lower Wilson Score. https://project-gc.com/Statistics/TopFavWilson

Awesome that they have this! Magnus is able to do some really cool things as they have special API access. :)

Link to comment
Share on other sites

  • 0
6 hours ago, Nic Hubbard said:

The problem with this is that it wouldn't be possible to do this with Live searching, since this uses the API and there is no API request or parameters that can do what you are asking.

I would assume that this means that weighted avg cannot be used for filtering as that is done by the API. But isn't sorting done client side? If that is the case, I would expect that this score can be added as a sort order. Furthermore, it can be added to the cache detail page, right? 

Adding weighted avg for sorting and cache detail page would already help me a lot!

Link to comment
Share on other sites

  • 0
1 hour ago, jasper said:

But isn't sorting done client side? If that is the case, I would expect that this score can be added as a sort order. Furthermore, it can be added to the cache detail page, right? 

Correct, it is.

I would be interesting in exploring this. 

12 hours ago, jasper said:

weighted rating = ( nr_favos + 3 ) / ( nr_premium_finds + 30 )

The only issue I see if this is that I cannot see how many finds of a cache are premium. Only how many finds in general.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...