Ratio vs. Raw Count

1.3k · January 21, 2011

Um. Last time I checked the information you want (specific real world cache examples) is an extension of the argument as to whether or not the information class of "ratio" should be available. That puts you in the camp of arguing about something that could be useful to others, but makes no difference to you if you won't use it.

Or am I missing something?

Yes you are missing something. I will clarify so there is no confusion.

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

2) What cache as 300 finds and 150 favorites.

What exactly am I missing there?

Neither of those have been answered or provided.

2.7k · January 21, 2011

Um. Last time I checked the information you want (specific real world cache examples) is an extension of the argument as to whether or not the information class of "ratio" should be available. That puts you in the camp of arguing about something that could be useful to others, but makes no difference to you if you won't use it.

Or am I missing something?

Yes you are missing something. I will clarify so there is no confusion.

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

2) What cache as 300 finds and 150 favorites.

What exactly am I missing there?

Neither of those have been answered or provided.

Right, so how does that invalidate what I said?

1.3k · January 21, 2011

Um. Last time I checked the information you want (specific real world cache examples) is an extension of the argument as to whether or not the information class of "ratio" should be available. That puts you in the camp of arguing about something that could be useful to others, but makes no difference to you if you won't use it.

Or am I missing something?

Yes you are missing something. I will clarify so there is no confusion.

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

2) What cache as 300 finds and 150 favorites.

What exactly am I missing there?

Neither of those have been answered or provided.

Right, so how does that invalidate what I said?

I doesn't. I was disagreeing with dfx, not you. I am not amazed at all this thread is still going on. I am just looking for answers.

7.2k · January 21, 2011

22k · January 21, 2011

Personally I'm amazed that this thread is still going. Why do people always argue against things that could be useful to others, but make no difference to themselves if they don't use it?

Fully agreed. I don't see what the problem is with providing this information to those who want it.

Am I supposed to simply let Groundspeak think I've got everything that I want and we can't do better?

5.7k · January 21, 2011

ungh...too much math...turning into a Gumby...

(not this one...

...this one)

2.7k · January 21, 2011

Personally I'm amazed that this thread is still going. Why do people always argue against things that could be useful to others, but make no difference to themselves if they don't use it?

Fully agreed. I don't see what the problem is with providing this information to those who want it.

Am I supposed to simply let Groundspeak think I've got everything that I want and we can't do better?

I'm not starting this mini-discussion up again. dfx made the point above, I agreed, then got into it with Keith. dfx can debate it further.

22k · January 21, 2011

...this one)

so in other words, your brain hurts.

Edited for clarification.

Edited January 21, 2011 by BlueDeuce

22k · January 21, 2011

I'm not starting this mini-discussion up again. dfx made the point above, I agreed, then got into it with Keith. dfx can debate it further.

Feel free to ignore the thread.

Edited twice for bad info on my part.

Edited January 21, 2011 by BlueDeuce

2.7k · January 21, 2011

I'm not starting this mini-discussion up again. dfx made the point above, I agreed, then got into it with Keith. dfx can debate it further.

Feel free to ignore the thread.

What, and miss the math primer?

(My brain hurts too, just looking at some of the math posts!)

1.4k · January 21, 2011

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

Here's the top 20 by Raw Count in Ontario. I don't get the "see if the math holds up" part as which result you want to use is up to personal preference.

Sort by Raw Count:

Sort by Ratio:

I personally like the ratio order.

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

2) What cache has 300 finds and 150 favorites?

Number 2 was answered in post #148.

Edited January 21, 2011 by Avernar

7k · January 21, 2011

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

Holds up against what? We're talking a purely subjective matter here, there is no absolute measure of cache quality or whatever else you hope to be able to read out of favorite counts. For some people the raw favorite count will work better for their purposes, for other people the ratio will work better, and for yet other people neither will work. Why do you want to deny people something that could be useful to them, just because you personally don't believe in it?

Am I supposed to simply let Groundspeak think I've got everything that I want and we can't do better?

Do you actually have a better suggestion? Or are you also just saying "it's of no use to me, so nobody else should have it"?

Edited January 21, 2011 by dfx

7.4k · January 21, 2011

If my PQ filter results could be ranked by favorites there might be some benefit.

Online, yes. Preview the PQ as a list and then sort on the favourite count column.

Offline, no. But hopefully soon.

That doesn't alway work. I have a PQ based from my home, 500 caches reach out about 10.9 miles. But if I sort on favorites in the preview, it takes the top 500 with favorites, some well over 100 miles away. I can't sort just the caches returned by that PQ. Some of the others I can.

If you mean your "Home" PQ, that's because you have it set up to go out 200 miles. 500 caches might only go out 10.9 miles, but when you sort on favorites, it's going to pull down the top favorites within that 200 miles. Try trimming down the radius on your query.

20.7k · January 21, 2011

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

Here's the top 20 by Raw Count in Ontario. I don't get the "see if the math holds up" part as which result you want to use is up to personal preference.

Sort by Raw Count:

Sort by Ratio:

I personally like the ratio order.

The problem is, no one can tell whether the math 'holds up' or not unless they go out and find those caches. Even then, they won't know if it did or didn't hold up simply because the individual's desires differed from the population being sampled.

1.3k · January 21, 2011

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

Here's the top 20 by Raw Count in Ontario. I don't get the "see if the math holds up" part as which result you want to use is up to personal preference.

Sort by Raw Count:

I personally like the ratio order.

Interesting. From what I can tell for the most part it reversed the list with a few anomalies which what it should should. From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on. I can now honestly say I prefer with the way people are assigning favorites, the RAW data.

2.7k · January 21, 2011

1) I have seen lots of examples sing made up numbers, see post #148. I would like someone to provide an example using 20 real caches and see if the math holds up.

Here's the top 20 by Raw Count in Ontario. I don't get the "see if the math holds up" part as which result you want to use is up to personal preference.

Sort by Raw Count:

I personally like the ratio order.

Interesting. From what I can tell for the most part it reversed the list with a few anomalies which what it should should. From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on. I can now honestly say I prefer with the way people are assigning favorites, the RAW data.

The other problem with the sample set is that the top 20 caches in Ontario by "RAW" are not the top 20 by ratio.

But that's impossible to sample on unless you work for Groundspeak at the moment so this is a nice little test. Looking at the list, and knowing which ones I have already found (most of the list) I would go by the "RATIO" method.

January 21, 2011

Sort by Raw Count:

Sort by Ratio:

I personally like the ratio order.

But you are not looking at a true ratio order for all Ontario caches.

You have posted the top 20 based on number, then resorted that limited sample based on ratio.

Where is the top 20 based on ratio alone?

How does that list compare to the top 20 based on number alone?

2.7k · January 21, 2011

Where is the top 20 based on ratio alone?

That's the functionality that some of us have asked for, and is being dismissed as not useful by others.

1.4k · January 21, 2011

Interesting. From what I can tell for the most part it reversed the list with a few anomalies which what it should should. I can now honestly say I prefer with the way people are assigning favorites, the RAW data.

I wouldn't call it reversed. The top did go to the bottom but the bottom really didn't go to the top. I did start with a list of top 20 by raw so maybe that caused that.

I'd really love to see the top 20 by ratio compared to the top 20 by raw but that requires help from Groundspeak or a lot of manual work.

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

No, people are placing favourites on everything on that list. What it shows is that I prefer physical containers and the harder to get to ones.

1.4k · January 21, 2011

But you are not looking at a true ratio order for all Ontario caches.

You have posted the top 20 based on number, then resorted that limited sample based on ratio.

Where is the top 20 based on ratio alone?

How does that list compare to the top 20 based on number alone?

Correct. I'd have to go through 140 caches, and that's with a raw threshold of 10 or more, and pick out the top twenty.

That's why we want the ratio to appear on the PQ preview page. A PQ filter option for minimum favourite would also be nice so we can exclude the low raw count ones because of sample size issues already discussed.

7k · January 21, 2011

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

Where do you deduct that from? How do you think that GCRDC1 could possibly get more favorite votes than GCMH1C?

1.3k · January 21, 2011

Interesting. From what I can tell for the most part it reversed the list with a few anomalies which what it should should. I can now honestly say I prefer with the way people are assigning favorites, the RAW data.

I wouldn't call it reversed. The top did go to the bottom but the bottom really didn't go to the top. I did start with a list of top 20 by raw so maybe that caused that.

I'd really love to see the top 20 by ratio compared to the top 20 by raw but that requires help from Groundspeak or a lot of manual work.

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

No, people are placing favorites on everything on that list. What it shows is that I prefer physical containers and the harder to get to ones.

Yes actually. Looking at the age and find count on the sample, there is a rough connection between favorites and finds with a few anomalies. If the first order is by favorites and the second order is based on favorites per find with favorites being connected to finds, then the result will be the inverse.

You prefer physical containers and the harder to get caches. The fact that the list agrees with you is coincidence considering the size of the sample.

8k · January 21, 2011

I'm happy to know that people want to use ratios to find caches they are more likely to enjoy (or as fizzymagic put it, "I just want to find caches that don't suck.") I was concerned that that an argument was being made the the cache with the higher ratio in a better cache. That would be just as bad as saying the cache with the higher raw count is a better cache. I think most people understand that certain caches are getting high raw counts because they get found more often. It doesn't seem to be as clear to some that certain caches will have higher ratios simply because these are found less often and generally by people who enjoy whatever the particular challenge is that makes that cache harder to find.

That said, I believe the reason that some are adamant about using the ratio is that they believe this number is a predictor of the the likelihood they will like a cache. I believe they are misinterpreting the information the ratio gives.

If you wanted to estimate the probability that a random geocacher would add a cache to their favorite list, I agree that a good estimator would be the percentage of previous finders that have added the cache to their favorite lists. You could of course use Markwell's method to get a better estimate: look at only premium members or only premium members who are using their lists. You could even go further and adjust for premium members who are not going back and favoriting caches they found in the past.

Ratios are an estimate of how likely a random cacher will favorite a cache. What we really want to estimate is how likely it is a particular geocacher (yourself) will enjoy a cache. Caches with high ratios are not necessarily cache that you will enjoy. These could be caches that involve solving a puzzle and you think that puzzle caches suck. Or it could be a hiking cache and hiking to a cache sucks. Or its a cache in an alligator infested swamp and you think that looking for caches in alligator infested swamps sucks. Now you might be able to eliminate sucking caches by some other means and then use the ratios on the rest, but does the higher ratio really mean that the you're more like to enjoy the cache? I think not. It may tell you (now that you've eliminated the cache you wouldn't have looked for anyhow) that you are more likely to add it to your favorite list. But it doesn't tell that if you don't add it to your list whether you would have enjoyed it or whether it thought it sucked. Of all the cachers who didn't put the cache on their favorites list we have no information as to whether they liked the cache but it just wasn't in the top 10% or if they hated the cache.

I contend that the raw count is more useful in finding caches that don't suck. The more geocachers that like the cache enough to favorite it (out of the entire population of geocachers) the more likely there is something about the cache that will appeal to a broad range of geocachers.

Now I'm the first to say I have my particularly likes and some of these are not things that a broad range of geocachers like. I like tough puzzles and I like long hikes. These caches will definitely score lower on raw counts than a virtual at a tourist location. But I can filter for these caches already and then look at the raw counts just in these caches. It could be argued that since the people who found these caches like caches similar to ones I like, I should look at the ratio here to see which cache I'm more likely to favor, But I'm probably going to enjoy any with a high raw count so it doesn't matter to me that it might be a fun cache but not one I will add to my favorites list. (This might matter more to someone else).

Now when I look at P&G caches that I might enjoy I can sort on the raw favorite count and visit a cache that many people have "recommended". Might I miss the new cache that every one who has found it so far liked because the raw count is lower. Sure. But it doesn't matter since I just want to find caches that don't suck. So what if I missed a new cache. There is always the possibility that the new cache has been found only by the hider's friends who all favorited it. The high ratio is no guarantee that the cache doesn't suck. Might the cache with lots of favorites suck. Well there is that chance, but I believe that if many people have put the cache on their favorites lists, the odds of the cache sucking are pretty slim. I'm going to stick with the raw count.

1.4k · January 21, 2011

Yes actually. Looking at the age and find count on the sample, there is a rough connection between favorites and finds with a few anomalies. If the first order is by favorites and the second order is based on favorites per find with favorites being connected to finds, then the result will be the inverse.

Your logic/math is faulty. If more people favourited the top caches on the first list they would have stayed put. Take a look at Mr Bean. That cache didn't really move.

The fact that the list agrees with you is coincidence considering the size of the sample.

I don't care if it's a coincidence or not. The list works for me, I will use the list. When the list stops working for me, I will stop using the list.

I'd love a bigger sample size! We need the sort option, or the find and favourite counts in the GPX files, to get the bigger sample size.

Edited January 21, 2011 by Avernar

1.3k · January 21, 2011

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

Where do you deduct that from? How do you think that GCRDC1 could possibly get more favorite votes than GCMH1C?

Oooooo, and easy one.

He likes the ratio results over raw. The ratio is opposite of raw. Raw is what people are placing favorites on. That means he like the opposite of what people are placing favorites on.

More people have found GCRDC1 to be a favorite then GCMH1C. Does that mean that is better? No. Does using a ration mean GCMH1C is better or more worthy of a visit? No. GCRDC1 simply has more favorites than GCMH1C.

I have never said one way was better than the other. In fact many including me have demonstrated previously that both are flawed.

1.1k · January 21, 2011

My opinion:

The ratio system is flawed for the following reason:

Only Premium Members (PM) can award favourites, hence the average cache as an example will be found by 30% premium members, 70% non premium members (I don't know what the exact percentage is...this is just a guess). However a PMO cache will be found on the most part only by PMs, hence a PMO cache will probably have a much higher ratio than a non PMO cache since 100% of finders of PMO caches can award a favourite but only 30% of finders can award a favourite to a non PMO cache. So for you math people...PMO and non PMO cache each has 100 finds and the caches are similar in quality and experience. 20 people each thought of this as a favourite. For the PMO cache, it would have 20 favourites, hence a 20% ratio, but the non PMO cache it would only have 6 favourites (only 30% of those 20 people can vote) for a 6% ratio. That is quite the spread and would easily skew the numbers when it comes to ratios.

Groundspeak decided to implement favourites as it is now. They didn't do the ratio system. Maybe that is something they will add in the future, or maybe they will see it as a flaw. If people want to take the time to figure out ratios, then good for them and they do so for their own interest sake. The way it is now, Favourites is interesting and it's something that adds some value\interest\feature, but that is all it is. For me it is not going to make me go out and find any specific cache because a bunch of people put a favourite on it, as some that I see in the top 20 I have to wonder how they could have so many such as the oldest active non-original, moved qualified cache in Ontario is #7 on the Ontario List and other than that qualification, there is nothing special or unique about it and is the same as thousands of other caches on the list and looking at the people who favourited it and reading their logs, they didn't have any type of unique experience, no wow factor, nothing and some even found the cache is horrible condition, but yet decided that a wet, smelly, moldy, musty cache deserves a favourite mark where if it was any other unqualified cache they would probably complain about the condition big time and put frownies in their logs.

So to me and to many I am sure, it's a neat feature but nothing more, nothing less. I'm not going out of my way to find these caches because they have a bunch of favourites on them, but instead I would go to a cache because someone says to me, "Hey Ralph, there's this cache that I think you will really enjoy visiting with your family. The cache\area is really neat and worth the time to visit it."

2.7k · January 21, 2011

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

Where do you deduct that from? How do you think that GCRDC1 could possibly get more favorite votes than GCMH1C?

Oooooo, and easy one.

He likes the ratio results over raw. The ratio is opposite of raw. Raw is what people are placing favorites on. That means he like the opposite of what people are placing favorites on.

How is ratio the opposite of raw. Isn't "no favorites" the opposite of raw? You're not going to immediately get an opposite list by switching to ratio - if it was that easy I'd just do a negative sort on raw and not ask for the ratio feature.

1.3k · January 21, 2011

From that and your statement of which you prefer I can tell you like the opposite of what people are placing favorites on.

Where do you deduct that from? How do you think that GCRDC1 could possibly get more favorite votes than GCMH1C?

Oooooo, and easy one.

He likes the ratio results over raw. The ratio is opposite of raw. Raw is what people are placing favorites on. That means he like the opposite of what people are placing favorites on.

How is ratio the opposite of raw. Isn't "no favorites" the opposite of raw? You're not going to immediately get an opposite list by switching to ratio - if it was that easy I'd just do a negative sort on raw and not ask for the ratio feature.

Already explained in a previous post. Look at the carefully. With the exception of a few cache, the list reversed.

20.7k · January 21, 2011

No, people are placing favourites on everything on that list. What it shows is that I prefer physical containers and the harder to get to ones.

So basically, you are attempting to make favorites approximate the results that you could much easier obtain through a PQ.

1.4k · January 21, 2011

He likes the ratio results over raw.

Agreed.

The ratio is opposite of raw.

Debatable.

Raw is what people are placing favorites on.

Bad english. You can't "place favourites on raw". The raw count is just that, a count. It can be increased by two things:

1) More visitors (objective)

2) A greater number of visitors giving it a point

If 1 goes up and 2 doesn't then the cache will drop on the list when sorted the other way.

That means he like the opposite of what people are placing favorites on.

Again, people gave favourites to all the caches on the list. Not enough of them placed favourites on some of the caches to keep them on top of both lists.

And finally, correlation does not imply causation.

7k · January 21, 2011

He likes the ratio results over raw. The ratio is opposite of raw. Raw is what people are placing favorites on. That means he like the opposite of what people are placing favorites on.

I think you're mixing up things now. The ratio is not the opposite (opposite would be either negative or reciprocal), but you may be thinking that the ratio is inversely proportional to the favorite count. But that would be true only of the favorite count was in the denominator of the ratio, which it isn't.

There's nothing "opposite" about it. If anything, the ratio would be the opposite of the find count (higher find count = lower ratio, with the same number of favorites).

Edited January 21, 2011 by dfx

1.4k · January 21, 2011

So basically, you are attempting to make favorites approximate the results that you could much easier obtain through a PQ.

I said prefer. I will hunt earthcaches and virtuals if they are interesting.

And even if I only wanted physical caches, how would I pick out the better ones? The ratio had the side effect of knocking a few of the non physicals further down but it still ordered the physical ones. And if the earthcaches or virtuals stayed on top (which they may have if we had the preferred denominator) I'd look at them more closely.

Besides, it's 20 caches. I'll get to all of them eventually.

5.3k · January 21, 2011

I was concerned that that an argument was being made the the cache with the higher ratio in a better cache. That would be just as bad as saying the cache with the higher raw count is a better cache.

While I disagree with you about whether the ratio or the raw is better, I completely agree here.

Caches with favorite points are generally better than those without. But it is ridiculous to say that cache A with 100 favorite points is absolutely better than cache B with 99. Nevertheless, I am absolutely certain that there will be cachers in need of ego boosting who will attempt to turn the whole thing into a competition. That is an unfortunate but inevitable side effect of any such system.

But whether you prefer ratios or raw counts, they are both vastly preferable to nothing, which is what we had before. I am thrilled to be able to use either. It's already made a difference for the better in my caching experience. And I think the current system, in which only positive feedback is possible, is the best design. Kudos to Groundspeak for adding this awesome feature!

1.4k · January 21, 2011

My opinion:

The ratio system is flawed for the following reason:

Only Premium Members (PM) can award favourites,

Yup! That's why we'd like the denominator of the ratio be the finders who have awarded at least one favourite point to any cache.

1.4k · January 21, 2011

I was concerned that that an argument was being made the the cache with the higher ratio in a better cache. That would be just as bad as saying the cache with the higher raw count is a better cache.

While I disagree with you about whether the ratio or the raw is better, I completely agree here.

Caches with favorite points are generally better than those without. But it is ridiculous to say that cache A with 100 favorite points is absolutely better than cache B with 99.

I agree. Favourite does not always mean better.

4.2k · January 21, 2011

I guess "close enough for jazz" just doesn't provide the requisite level of detail you guys are looking for, huh?

2.2k · January 21, 2011

It's the objects that make it impossible to compare the two, not the methodology. People are trying to compare tourist caches to long walk in the woods caches. They surmise that using a ratio will instead of raw data will make this comparison possible, but it won't. You still have to target the kinds of caches that you like and look at the fave data for only that subset.

That's exactly my thought.

Here's what I know. I create a PQ centered over my house and preview it. I sort that preview by the number of Favorites.

Guess what? The caches with the most Favorites are the ones that I indeed think are more enjoyable than the others.

I like the comparison to "off the beaten path" caches with "tourist" caches.

Do peoples' standards really drop when they are on vacation? When I am away from home, it suddenly doesn't make the skirt lifter in my hotel parking lot any more enjoyable than it would be at home. I may be more likely to do it than I am at home, but being away from my home coordinates doesn't suddenly make that cache more "Favorite-worthy" than it would be normally.

1.8k · January 22, 2011

I guess "close enough for jazz" just doesn't provide the requisite level of detail you guys are looking for, huh?

I'd find it useful.

5.6k · January 22, 2011

If my PQ filter results could be ranked by favorites there might be some benefit.

Online, yes. Preview the PQ as a list and then sort on the favourite count column.

Offline, no. But hopefully soon.

That doesn't alway work. I have a PQ based from my home, 500 caches reach out about 10.9 miles. But if I sort on favorites in the preview, it takes the top 500 with favorites, some well over 100 miles away. I can't sort just the caches returned by that PQ. Some of the others I can.

Ahh! Now I understand. I was expecting the list to stay static and just the order to change. But sorting on faves changes the selection. Interesting. Not always helpful, but interesting.

If you mean your "Home" PQ, that's because you have it set up to go out 200 miles. 500 caches might only go out 10.9 miles, but when you sort on favorites, it's going to pull down the top favorites within that 200 miles. Try trimming down the radius on your query.

20.7k · January 22, 2011

So basically, you are attempting to make favorites approximate the results that you could much easier obtain through a PQ.

I said prefer. I will hunt earthcaches and virtuals if they are interesting.

And even if I only wanted physical caches, how would I pick out the better ones? The ratio had the side effect of knocking a few of the non physicals further down but it still ordered the physical ones. And if the earthcaches or virtuals stayed on top (which they may have if we had the preferred denominator) I'd look at them more closely.

Besides, it's 20 caches. I'll get to all of them eventually.

You pick out the better ones using the tools available, one of which is the favorite count, but the tools are not limited to that. What I get from your posts is not that you are looking to pick out the better caches to find (ie good caches vs not good caches) but that you are trying to make favorites alone show you the absolute best caches. This is not a job that faves can perform, in my opinion due to all of the reasons for error and skew already discussed in this thread and the various threads which have discussed this topic previously.

8k · January 22, 2011

Besides, it's 20 caches. I'll get to all of them eventually.

Why then would you care to order them is a way you think better reflects the enjoyment (or the odds of enjoyment) you would have. Must you find the "best" cache today and one that is second "best" tomorrow?. Would it hurt to look of a cache that you have a good chance of enjoying today and leave the one you have an even better chance of enjoying till tomorrow?

Of course where the favorites will have the most usefulness is when you travel someplace. You know you won't have time to find all the caches there so you do want to maximize the probability that you enjoy those caches you search for. Then the favorites becomes one more thing you can use in selecting caches. If you believe that the ratio better predicts how likely it is you will enjoy the cache, you can use it that way. But even you have indicated that you can't use the ratio that way. Instead you look at the raw count and when it is low you discount the ratio. Only if the raw count is above a threshold do you then use the ratio to get a finer resolution on which caches you are more likely to enjoy. Given all the other factors, I can't imagine that the ratio truly adds any information. When I use the raw count it's true that I will likely look to see if the favorite is high solely because the cache has lots of finders. However, if I'm somewhere on vacation, just the fact that a cache gets lots of finder would already interest me. It may be a tourist destination that I am thinking about visiting anyhow. The fact that there is a cache there that a lot of people have favorited may be another reason for me to go there. This could very well be more important in selecting which caches I will look for than what percentage of the finders liked the cache. In fact I can't think of a reason that the ratio would make a difference. (The underlying reasons for a high/low ratio may make a difference, but I don't need a ratio to determine this).

1.8k · January 22, 2011

Given all the other factors, I can't imagine that the ratio truly adds any information... In fact I can't think of a reason that the ratio would make a difference.

I find that it adds information. When we travel as a family I am often given a hall pass to pick one or two caches to mark the visit on my map. In those situations I am often particularly uninterested in simply finding the caches that have been found most often by tourists thanks to their easy location. I may not want my one caching memory to be a small container near hotels, that only 2% of visitors voted for.

My own preferred list would probably be filtered for caches with at least 50 finds, and then ordered by percentage favorite votes. I believe that list will be more efficiently sorted for me, as I try to pick which cache or two to spend my chits on.

1.3k · January 22, 2011

I went down East three years ago and then out West last year. There were no such thing as favorites so I had to do some research. Turns out that I hit the nail onthe head as those turn out to be the ones with the most favorites. This tells me that RAW will work for me.

2.7k · January 22, 2011

Right Keith. So rather than ratios, you would suggest we just get rid of favorites altogether, as they are unnecessary.

Back in 2003 I didn't have paperless Geocaching in the field- does not mean I find it useless today. Same will happen with favorites.

1.8k · January 22, 2011

This tells me that RAW will work for me.

I would never suggest that raw totals wouldn't work for you.

I am less clear about the insistence that ratio numbers wouldn't be useful for me.

1.3k · January 22, 2011

Right Keith. So rather than ratios, you would suggest we just get rid of favorites altogether, as they are unnecessary.

Back in 2003 I didn't have paperless Geocaching in the field- does not mean I find it useless today. Same will happen with favorites.

Never suggested that at all. In fact I stated RAW will work for me. Kind of gives me a warm fuzzy feeling that I am not missing out on what other people think are one of their favorite.

8k · January 22, 2011

Given all the other factors, I can't imagine that the ratio truly adds any information... In fact I can't think of a reason that the ratio would make a difference.

I find that it adds information. When we travel as a family I am often given a hall pass to pick one or two caches to mark the visit on my map. In those situations I am often particularly uninterested in simply finding the caches that have been found most often by tourists thanks to their easy location. I may not want my one caching memory to be a small container near hotels, that only 2% of visitors voted for.

My own preferred list would probably be filtered for caches with at least 50 finds, and then ordered by percentage favorite votes. I believe that list will be more efficiently sorted for me, as I try to pick which cache or two to spend my chits on.

Great, if that works for you. I wonder though how you filter for caches with at least 50 finds as there is no PQ option to do so. Even in GSAK you can't do this, since the PQ only shows the last five logs. My guess is that you might be looking at the raw favorites counts and then look at ratios on the caches that have the most favorites. Then you assume that those with a low ratio but lots of favorites are caches you're not interested in looking for. And that's OK. It may very well be that these are always virtuals in touristy locations or have a high number of finders for some reason that doesn't appeal to you.

It's clear that there is a correlation between number of finds and the raw favorite count and that people would like to account for this in order to seek caches that are found less often but may be just as enjoyable or perhaps even more enjoyable. I'm not convinced that ratio does this. It's assuming the correlation between find count and raw number is very strong and linear. My guess is that some of the caches that are found most often get a lower ratio for other reason than just number of finds. We already have discussed that these are older caches, and that some premium members are not going back and awarding favorites to caches they found a long time ago. Right there is a reason these caches have a lower ratio. But beyond this, a easy grab in a touristy location may be attracting more geocachers who do not use the favorites at all. There may be a higher percentage of non-premium members who find these caches. Of course if you computed a ratio against only premium member finders who use the favorites system and have ether found the cache since favorites came out or have gone back and awarded favorites to caches they found a long time ago, you could eliminate these reasons for a lower ratio. I doubt Groundspeak will compute ratios this way. In any case, I could probably come up with other reasons why ratios are lower on some caches that get found a lot. An obvious on is that cachers see the cache already is getting a lot of favorites so they decide to give their vote to another cache. (It's really hard to decide how to award those last few favorite votes so why not use them to recommend a cache in your top 11% if one in the top 10% is already getting a lot of recommendations from someone else.)

Now I'm not saying that raw counts aren't affected by these same problems, but I find that the supposed extra information in the ratio is not really there. In either case you need to look at the individual cache and decide whether you want to look for it or not.

7k · January 22, 2011

Never suggested that at all. In fact I stated RAW will work for me. Kind of gives me a warm fuzzy feeling that I am not missing out on what other people think are one of their favorite.

What's one got to do with the other? Even if you'd see only the ratio, you could (probably) still click on the bagde and see whose favorite it is.

1.8k · January 22, 2011

My guess is that some of the caches that are found most often get a lower ratio for other reason than just number of finds.

Caches don't have a low ratio *because* they have a lot of finds. Not at all. It's generally because the caches themselves aren't interesting enough to get votes from more than (say) 2% of the cachers who find them. (Yes, it's true that there are weird effects for a few special classes of caches, like really old caches where none of the finders have logged on in years, or caches that for some reason are never found by premium members... I'm holding those out as special exceptions for the moment.)

If there is a cache out there with 200 finds and 50 favorites votes, I certainly don't expect the ratio to drop significantly as the cache approaches 500 or 1000 finds.

But to be clear, I completely agree that there are a lot of reasons that caches get higher or lower ratios, that have *nothing* to do with how often they are found. And that's exactly why I'd love to sort on ratios.

In either case you need to look at the individual cache and decide whether you want to look for it or not.

Well, yes. I would think that would apply to every metric in the database.

1.3k · January 22, 2011

Never suggested that at all. In fact I stated RAW will work for me. Kind of gives me a warm fuzzy feeling that I am not missing out on what other people think are one of their favorite.

What's one got to do with the other? Even if you'd see only the ratio, you could (probably) still click on the bagde and see whose favorite it is.

One and other what? I don't understand then question?

Ratio vs. Raw Count

Recommended Posts

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment