Calculation of FP % is incorrect

[+Goldenwattle]

On 1/5/2024 at 3:44 AM, LydiaSimmons said:

Numerator is displayed accurately on the top of the cache page.  Total number of finds is displayed accurately on the bottom of the cache page. I'm just saying, there's enough information on the page to calculate a FP percentage.  Premium logs is not well-defined anyways as memberships get renewed, and lapse, etc.

Well , that will take some time when listing hundreds of caches for a trip. Not practical.

 

Percentage is MUCH more important than number to judge a cache by. If people think that there is enough information already (I don't agree), then just have the percentage of favourite points instead, as that gives a far better idea of the cache quality. In some places caches are found almost every day. In Europe I found some caches that were found this often; in fact it wasn't uncommon for some to have several finds in a day. In other places a cache might only have one or two finds (maybe even less) a year. So someone comparing these caches, says look this one (with several thousand finds) has 200 favourites so it must be the best cache. In reality it only has 10% favourites, but most people won't want to waste time working this out. Most people haven't thought it worthy of a favourite. Meanwhile the other cache has a piddly 5 favourites, so in comparison it can't be a good can it. I mean only five favourites . That's nothing! In fact it has 100% favourites.

As in many things in life percentage gives the truer picture.

[+baer2006]

8 hours ago, Goldenwattle said:

Percentage is MUCH more important than number to judge a cache by. If people think that there is enough information already (I don't agree), then just have the percentage of favourite points instead, as that gives a far better idea of the cache quality. In some places caches are found almost every day. In Europe I found some caches that were found this often; in fact it wasn't uncommon for some to have several finds in a day. In other places a cache might only have one or two finds (maybe even less) a year. So someone comparing these caches, says look this one (with several thousand finds) has 200 favourites so it must be the best cache. In reality it only has 10% favourites, but most people won't want to waste time working this out. Most people haven't thought it worthy of a favourite. Meanwhile the other cache has a piddly 5 favourites, so in comparison it can't be a good can it. I mean only five favourites . That's nothing! In fact it has 100% favourites.

As in many things in life percentage gives the truer picture.

I agree 100%!

 

In fact, only seeing the absolute number of FP can also be somewhat embarrassing as a cache owner. My cache with most FP has 147 - this looks quite ok, but the cache has more than 18000 finds! So the FP percentage is 0.8%, and even that is arguably 0.8% too high (it's a plain D1/T1 film canister - no idea, why people give FP for that. Maybe some like the easy find). OTOH, a few of my caches have quite a decent FP percentage (not the "top league", but > 30%), but the absolute number is unimpressive, because they don't get many finds (for a cache not far from a large city in Germany).

[+Goldenwattle]

34 minutes ago, baer2006 said:

I agree 100%!

 

In fact, only seeing the absolute number of FP can also be somewhat embarrassing as a cache owner. My cache with most FP has 147 - this looks quite ok, but the cache has more than 18000 finds! So the FP percentage is 0.8%, and even that is arguably 0.8% too high (it's a plain D1/T1 film canister - no idea, why people give FP for that. Maybe some like the easy find).

That's the sort of cache I was referring to I found in Europe. Extremely ordinary, but with a large number of FPs.

 

My cache (GC847R1) with the highest FPs has 83 favourites, for 242 finds. It's a furnished and fitted out TB hotel. Carpet, wallpaper, paintings on the wall, beds and other furniture, tiny suitcases, etc. Even has a working wall clock. No insult intended (I too also have other caches similar to the cache you described), but only giving FPs is INSULTING to geocachers who have made the extra effort to precent something more interesting. Which is going to attract more people to stay in geocaching; another film canister or bison (my boring caches), or an interesting, thought out cache?

 

My two other caches with the highest percentage of FPs are:

Guarded (yes, it's in a guard, but the name doesn't only refer to the road guard) GC7AWYW, published 2017, with 26 FPs for 79 finds.

Red Hill GC4CB12 (published 2013), with 37 FPs for 119 finds.

 

They demonstrate why number of FPs is not an accurate measure of how good a cache is. Your cache has 18,000 finds . I can't even imagine that here in Australia. My Red Hill cache for instance, published in 2013 has only 119 finds. That's only about 12 finds per year on average. And the cache is in a suburb, not out in the country.

 

Giving only FPs is not comparing apples with apples.

Edited January 21 by Goldenwattle

[+fizzymagic]

Favorite point percentage is # of favorites divided by number of premium user logs.

 

Non premium users can't leave favorite points even if they want to.

 

If that bothers you a lot then make all your hides premium-only.

Edited January 22 by fizzymagic

[+Goldenwattle]

43 minutes ago, fizzymagic said:

Favorite point percentage is # of favorites divided by number of premium user logs.

I didn't know that. My caches I mentioned as examples, are Premium User only caches.

 

The % is not visible though without clicking on it.

[+LydiaSimmons]

13 hours ago, Goldenwattle said:

Well , that will take some time when listing hundreds of caches for a trip. Not practical.

 

Percentage is MUCH more important than number to judge a cache by. If people think that there is enough information already (I don't agree), then just have the percentage of favourite points instead, as that gives a far better idea of the cache quality. In some places caches are found almost every day. In Europe I found some caches that were found this often; in fact it wasn't uncommon for some to have several finds in a day. In other places a cache might only have one or two finds (maybe even less) a year. So someone comparing these caches, says look this one (with several thousand finds) has 200 favourites so it must be the best cache. In reality it only has 10% favourites, but most people won't want to waste time working this out. Most people haven't thought it worthy of a favourite. Meanwhile the other cache has a piddly 5 favourites, so in comparison it can't be a good can it. I mean only five favourites . That's nothing! In fact it has 100% favourites.

As in many things in life percentage gives the truer picture.

 

I was not suggesting to omit the percentage. I was suggesting a change to how the percentage is calculated.  I propose to change the denominator to not use the number of premium logs, but instead the number of total found it logs.  I suspect it is computationally expensive to look up the current membership status of each finder of a cache when the % score is requested, and so some data is being cached, which adds complexity to this system.  

Edited January 22 by LydiaSimmons

[+Goldenwattle]

31 minutes ago, LydiaSimmons said:

I was not suggesting to omit the percentage.

I never suggested this .

[+fizzymagic]

2 hours ago, LydiaSimmons said:

I was not suggesting to omit the percentage. I was suggesting a change to how the percentage is calculated.  I propose to change the denominator to not use the number of premium logs, but instead the number of total found it logs.  I suspect it is computationally expensive to look up the current membership status of each finder of a cache when the % score is requested, and so some data is being cached, which adds complexity to this system.  

 

So having a percentage based on an unknown distribution makes some kind of sense?  There is no way to tell, based on total logs, what percentage of those people could have given it favorite points.  So the proposed metric includes less information than the current one, uses hidden information, and is open to gaming.  I think it is not a good idea.

[+LydiaSimmons]

8 hours ago, fizzymagic said:

 

So having a percentage based on an unknown distribution makes some kind of sense?  There is no way to tell, based on total logs, what percentage of those people could have given it favorite points.  So the proposed metric includes less information than the current one, uses hidden information, and is open to gaming.  I think it is not a good idea.

 

How do you measure whether someone could have given the cache a favorite point?

 

If they were a basic member at the time they logged a cache, but then purchased a premium membership years later, they could go back and favorite the cache.  But if they don't. Should their newfound premium status now make their log count as premium? Only if they became a premium member for the first time and were granted favorite points for their past finds?

 

Similarly, if someone was a premium member when they logged a cache, and didn't give it a favorite point.  And then let their membership lapse.  Does their found it now count as a non-premium found it? That's not really accurate either; they didn't favorite it when they could.

 

Whether you measure the finder's current premium status or the premium status at the time of logging, # of premium logs is subject to flaws.  Using the total number of finds is simpler, more stable, and I believe, good enough to get a sense of how people are enjoying the cache.  The CO could make the cache premium members only if they wanted, as well (yes, I know there is a way for a basic member to log a premium member only cache).

[+CCFwasG]

On 1/4/2024 at 11:44 AM, LydiaSimmons said:

Numerator is displayed accurately on the top of the cache page.  Total number of finds is displayed accurately on the bottom of the cache page. I'm just saying, there's enough information on the page to calculate a FP percentage.  Premium logs is not well-defined anyways as memberships get renewed, and lapse, etc.

 

On 1/21/2024 at 8:39 AM, Goldenwattle said:

Well , that will take some time when listing hundreds of caches for a trip. Not practical.

 

 

My take on the above is that the replies misunderstood the post. No one suggested that cachers do the calcs themselves, I think, just that there's enough information on the page FOR Groundspeak to do the calcs and get it right. (Although I will suggest it in a sec. Listing caches for a trip I don't think shows % only FPs? Unless using an outside website I guess?)

 

1 hour ago, LydiaSimmons said:

 

How do you measure whether someone could have given the cache a favorite point? 

... [clip]

Whether you measure the finder's current premium status or the premium status at the time of logging, # of premium logs is subject to flaws.  Using the total number of finds is simpler, more stable, and I believe, good enough to get a sense of how people are enjoying the cache.  The CO could make the cache premium members only if they wanted, as well (yes, I know there is a way for a basic member to log a premium member only cache).

While I sort of agree here, the huge dip in %FPs for those caches that are basic and getting huge numbers of finds but maybe more from basic members than premium, the calc would be really unfair to the CO/cache. 

Here is an odd example btw:  a cache with over 200 FPs and 8900+ finds...  https://coord.info/GC26VRH
Currently it says 5% FPs, but the actual calc if you include all logs is 2.7% (half!). So I suspect (but obvs am not going to count) it is calculating the % from the PM logs.  And granted, people go to it because they see something with a LOT of FPs. And yet actually it has a low percentage. (And isn't that great of a cache to be honest!) My point is, I think, that there is no winning here. Nothing suggested by anyone is "perfect" and actually all are problematic. I understand that FPs can be valuable, but they are often over-hyped and over-used IMHO. I don't want to banish them, but honestly I would just get rid of the percentage thing. Let people look and figure from the FPs whether to go for something, or look at FPs vs number of finds, if so inclined. I think the % calc has made a lot of work for GS programmers with results that are unsatisfactory to most everyone in some way. Not to mention buggy as h*ll!

 

Edited January 22 by CCFwasG
correction

[+thebruce0]

Here's a way it could happen to be up to date, and I don't think FPs are stored this way:

* Post a find log, and if PM, choose to add a FP to the log.

* No longer a PM? Your FPs are still associated with your find logs (you can remove, but you can no longer add)

* Find a cache, then want to add a FP? Go to the listing, and add it (only an option if you're PM and already found it), which associates the FP to the log.

 

Metrics:

1. Total FPs added to all logs (current PM or not doesn't matter)

2a. Filter out non-current-PM-logs

2b. Re-count FPs (now current PM logs), over total PM logs = Current PM percentage.

 

There's no way as mentioned to determine who could have given an FP and didn't on past find logs without hefty date-based filtering, so the percentage can never be truly accurate as a general count. So make it the context of current PM only. Yet we still want to see total FPs.

- Total FP count: #

- Current PM FP percent: PM-FP/PM-Finds %

 

And tying the FP to the log, not as a user-to-cache data point, means that Find logs can be deleted and the above numbers will still remain accurate. The only PM-check will be for the current percentage when the log count is reduced to current PM-only users.

 

If the count and % are calculated live, no issues.

If the numbers are stored asynchronously (counts and % stored, not calculated live with every view), then passively trigger the FP recalculation on:

* Add/Remove FP

* Post/Edit/Delete Found It log

* And (depending on available cycles) on serving a cache listing detail (web or mobile) where a Found It logger's PM status has changed (may also require a new data point if not already existing)

* Edit listing (for COs who really want to force-update the FP status if it seems inaccurate)

 

Edited January 22 by thebruce0

[+Zyklotrop]

Hi!
I have a cache. Fine.
It is GCADECW
It had yesterday 8 founds with 8 Fav Points and a quote with 100%.
So far so good.
Today it has 9 Founds with 9 Fav Points and a quote with 89%.
Not so good.
All Geocachers that had found the cache are Premium.
Why is the quote not 100% anymore?

 

Greetings 
Zyklotrop

[Keystone]

The above post was originally a separate thread.  I merged it into this existing discussion.

[+2Abendsegler]

On 5/18/2023 at 8:22 PM, worrellsquirrel said:

This issue is still being investigated by our engineering team. We apologize for any inconvenience but thank you for your patience in the meantime.

Hi @worrellsquirrel,

 

I have another fresh example of the incorrect calculation of the favorites percentage. The new cache so far only has one single found log.

The cacher is a premium member and has set a favorite point when logging with the new log form. The cache also has this favorite point. However, <1% favorites are shown as favorites percentage.

Can you perhaps address the problem again with the development team?
Do you need more examples or is there anything else I can contribute to the solution?

[+2Abendsegler]

On 5/18/2023 at 8:22 PM, worrellsquirrel said:

This issue is still being investigated by our engineering team. We apologize for any inconvenience but thank you for your patience in the meantime.

 

Hi,

 

I have another fresh example of the incorrect calculation of the favorites percentage. Our cache so far only has since today two found logs.

Both cacher are premium members and have set a favorite point when logging with the new log form. The cache also has this favorite points. However, <1% favorites are shown as favorites percentage.

 

 

Can you perhaps address the problem again with the development team?
Do you need more examples or is there anything else I can contribute to the solution?

 

[+zuza88]

On 4/8/2023 at 8:32 PM, baer2006 said:

Before attempting to "fix" anything here, one must first establish a rigorous definition of the metric "percentage of FP". And I don't think that's super easy.

Basically, it should be "#FP / #(Logs from PMs)", because that's what it says on the listing page: xx% Favorites/Premium Logs

 

A) The simple approach would be #FP/(# of finders, who are currently PM). But this can easily lead to totally misleading percentages, because when a PM, who has given an FP, becomes BM (e.g. because the quit the game, and let the premium membership expire), the FP remains. Example: A cache with 10 finds, 6 FP, but only 2 finders are PM (because 4 of the those with an FP are now BM). So what's the percentage now? #FP/#PM-finders would give 300%, which doesn't make sense.

 

A better approach would be (#FP given by current PMs)/(# of finders, who are currently PM). With this definition, the above example would give 100% - the two remaining PM finders both gave an FP. How many ex-PMs gave an FP in the past would be irrelevant.

 

C) One could even define it like this: #FP/(#FP + # of finders who are currently PM and did not give an FP). Sounds complicated, but boils down to counting in the denominator all finders, where we know they could have given an FP: Those who actually gave one plus those who are currently PM and therefore could give one.

 

Both B and C should be easily query-able from the database, using only persistent information (i.e. who has found the cache, who is currently PM, and who has given an FP).

Recently I've been observing a case when there was 0 find logs, and 1 FP given (sic!). I've seen it twice. How that could have happened? The cache was published, some cacher logged it and gave a FP, then the owner deleted a log - yes, in this scenario the FP sticks to the cache! So, we now have 0 finds and 1 FP. Moreover, no one can do anything - neither the cacher who originally logged it (there is no "Remove FP" option anymore, since there is no log), nor the cache owner (he can't remove given FP in any way). Now, in this scenario, what should be the % of FP given? The answer is...: infinity

 

Of course that's another, related bug, since when the owner removes a log with FP, the FP should be returned back to the cacher.

 

I'm not sure what happens when the cacher removes his log with a FP given before - will FP stay there, or will it be automatically returned to him?

 

And I guess even with this bug, the option C would still work the best. However, if only basic members have found a cache, we are geting 0/0, what should be interpreted as 0%.

[+CocoBerlin]

Hey HQ team,
 

Looks like this hasn't been fixed at all yet.

Another example:

 

My cache GCANMYC has 25 FPs / 29 PMs (of 31 total finds) = 80% FPs (should be 86%) 

 

@HQ can you please give us an update? Will you ever work on this issue?

[+Viajero Perdido]

Birthday isn't until September.

 

How about switching to a simple one-item bar graph without a scale?  Fewer people, I suspect, would hold a ruler to the screen to check the math.

[+2Abendsegler]

3 hours ago, CocoBerlin said:

My cache GCANMYC has 25 FPs / 29 PMs (of 31 total finds) = 80% FPs (should be 86%) 

Perhaps your calculation is too simple. A Premium Member is someone who is a Premium Member at the time of the log and not at the present time. For example, the cacher "kornecke" has set a favorite point, but is now only a basic member.

 

But "Team CASA87 sagt DANKE und gerne vergeben wir hier einen Favoriten", but has not set one. Perhaps they could be reminded of this.

[+Viajero Perdido]

57 minutes ago, 2Abendsegler said:

But "Team CASA87 sagt DANKE und gerne vergeben wir hier einen Favoriten", but has not set one. Perhaps they could be reminded of this.

 

(said THANKS but hasn't...)

 

That's finesse.  But shouldn't basic math come first?  Adding a favorite point should never - in this universe and its rules of math - instantly lower the favorites-percent.  Spock would never land on this planet.

 

I keep apologizing when I give favorite points, which has become a back-handed compliment these days...

 

Groundspeak, please explain how to give a favo point and increase the percentage at the same time.  Surely there's a code path that you've tested ... what is it?  Thanks.

 

Edited September 1 by Viajero Perdido

[+barefootjeff]

6 hours ago, 2Abendsegler said:

Perhaps your calculation is too simple. A Premium Member is someone who is a Premium Member at the time of the log and not at the present time.

 

Someone who was a basic member at the time they found a cache but subsequently becomes a premium member can always go back and add an FP to it, which can complicate the calculation. But lest there be any doubt about the bug, one of my caches, placed in October last year, has four finders and received three FPs. All of those finders were PMs at the time they found it and are still PMs, so I would have thought the percentage FPs would be a pretty unambigous 75%. HQ seem to use a different sort of arithmetic though:

[+2Abendsegler]

18 minutes ago, barefootjeff said:

Someone who was a basic member at the time they found a cache but subsequently becomes a premium member can always go back and add an FP to it, which can complicate the calculation.

Yes, that's true.

 

18 minutes ago, barefootjeff said:

But lest there be any doubt about the bug

I have no doubts either, I know caches myself where the calculation follows a different arithmetic.

[+2Abendsegler]

6 hours ago, Viajero Perdido said:

But shouldn't basic math come first?  Adding a favorite point should never - in this universe and its rules of math - instantly lower the favorites-percent.  Spock would never land on this planet.

Of course. 😂

But that was just an example that it might be more complicated.

 

And it seems less finesse to me. It seems more like it was forgotten.

[+MrDosinger]

So, HQ, once again: what are you planning to do about this?