Weighted Cache Find Count

[Hedgehog in the Fog]

I have tried to research this before posting but did not find anything sufficiently similar in the recent history of the forum.

 

I am not a numbers person (in the sense of being after the high count) but I do like the numbers to be meaningful if they are used. Obviously the effort that goes into 5/5 cache is incomparable to 1/1. The person with 100 finds of above average difficulty rating is likely spend more time and effort than 200 finds of easier nature.

 

Have the weighted count being considered as an official site feature or a stats generator feature? I would propose something like this (while retaining two numbers, for the hide difficulty and terrain difficulty):

 

exponential weighted raiting = (2^star)/8 which would give us

 

1 star => 0.25

2 star => 0.50

3 star => 1.00

4 star => 2.00

5 star => 4.00

 

The sum of the weighted ratings would provide the weighted finds count better approximating total amount of effort and experience.

 

Comments? If nothing else it would be great to have such or similar weighted counted implemented in some stats generator. If there is enough interest I might play with some coding myself even though this is not my main strength.

[+M 5]

Some stat generators have stats that show you average D/T rating or the percentage of find over a 3 in either Difficulty or Terrain. I haven't seen anything weighted. I would be interested in something along those lines. Most of the numbers on your profile page are just for you and your friends anyway. Unless some challenge caches take generated stats into account. I know that Groundspeak is currently looking into an official stat generator. Not sure the status on that or what issues they are looking into. You may want to put the idea up on the feedback site.

[+addisonbr]

I understand where you're going with it, but I think the math is more complex than people would be generally comfortable with. When you start using power functions, you're going to end up with numbers that lose the benefits of a straight-forward, real-world explanation. If someone told me that she had 1257 points by your system, I would have a really hard time understanding what that meant.

 

If you wanted there to be traction with your calculation, you'd probably have more luck with something simpler - even as simple as just adding up stars. "Stars collected" probably accomplishes 80-90% of what you're trying to quantify, and while it is assuredly still flawed, it would retain some intuitive connection between your number and the real world.

[+dfx]

there's similar systems already out there, but i can't think of any of their names right now. none of them really caught on i guess.

[+Xaa]

You can't even compare various caches with the same rating. People in Switzerland will laugh at the 3 stars for terrain some caches in the Netherlands get.

I think your suggestion is as good a suggestion as it would be to use a weight factor of 1 (in other words: the way it is now).

If you want to compare with another cache, feel free to use whatever criterium you like, but I guess it's not really possible to come up with a decent comparison between any pair of cachers.

The only real way might be to compare the amount of fun had, and I guess that's hard to catch in a formula.

[+dfx]

You can't even compare various caches with the same rating. People in Switzerland will laugh at the 3 stars for terrain some caches in the Netherlands get.

very true!

[+niraD]

One of the local cachers has a personal goal of maintaining an average terrain rating of at least 2.0, but that's a straight average, rather than a weighted average like you've described.

[+StarBrand]

It still won't measure the fun I had hunting them.

[+dakboy]

mdCachingPoints

 

Generated automatically by GCStatistic. I think there's a GSAK macro that calculates it as well; if not, I wrote one (and in the process discovered a bug in GCStatistic's calculation which has since been fixed).

[+sbell111]

I have tried to research this before posting but did not find anything sufficiently similar in the recent history of the forum.

 

I am not a numbers person (in the sense of being after the high count) but I do like the numbers to be meaningful if they are used. Obviously the effort that goes into 5/5 cache is incomparable to 1/1. The person with 100 finds of above average difficulty rating is likely spend more time and effort than 200 finds of easier nature.

 

Have the weighted count being considered as an official site feature or a stats generator feature? I would propose something like this (while retaining two numbers, for the hide difficulty and terrain difficulty):

 

exponential weighted raiting = (2^star)/8 which would give us

 

1 star => 0.25

2 star => 0.50

3 star => 1.00

4 star => 2.00

5 star => 4.00

 

The sum of the weighted ratings would provide the weighted finds count better approximating total amount of effort and experience.

 

Comments? If nothing else it would be great to have such or similar weighted counted implemented in some stats generator. If there is enough interest I might play with some coding myself even though this is not my main strength.

Wouldn't it be simpler to just give one point for each 'star'?

 

1 star => 1

2 star => 2

3 star => 3

4 star => 4

5 star => 5

[Hedgehog in the Fog]

I understand where you're going with it, but I think the math is more complex than people would be generally comfortable with. When you start using power functions, you're going to end up with numbers that lose the benefits of a straight-forward, real-world explanation. If someone told me that she had 1257 points by your system, I would have a really hard time understanding what that meant.

 

Math would be done by software - I would not advocate calculating the points by hand :-) I agree that the power law is not straight forward but I was interested in something that reflects on the non-linearly increasing challenge of finding 5 star vs 1 start cache (not 2 min vs 10 min but more like 2 min vs 10 hrs).

 

All the users would need to understand is that 3 stars corresponds to an average "1 find", 1 or 2 stars give you a 1/4 or a 1/2 of "a find", 4 stars and 5 stars five you 2 and 4 "finds" respectively.

 

I agree with most comments above: such statistics have to be taken with a grain of salt since not all 3 star caches are the same in difficulty. More importantly how to you measure fun?? I don't want to (even though temped) to delve into the proposal of cache rating by the seekers rather than the owner - seems to come up here at regular intervals :-)

 

I understand that user stats are mostly for the user (I don't look at geocaching as a competetive sport) and the friends and that is also why I think it is still possible to discuss the stats as one's own measure of progress without worrying too much about the rating errors.

[+Harry Dolphin]

Seems like the dumbest idea I've seen in quite a while. Apologize for that.

It would seem to be a meaningless number. Precision versus accuracy. Without accuracy, precision is meaningless. Everyone knows that 4.5 is much harder than 5! 5s tend to be overrated, or just require special equipment. (A good winter freeze can change that.) 1s tend to be underrated. Of the four 5 terrain caches that I've done, only the 1/5 was tough terrain. We hiked three days, with 4000' of climb to get to it, true. But it could be done with 1000' of climb. And the 1 difficulty was underrated! Took us a half hour to find it! I would probably have called it a 2/4. Ratings are a subjective guideline. So, outside of a pretty number with little meaning, I have no idea what porpoise such a number serves. But, if it floats your boat, go for it! But do not expect anyone else to attach any meaning to it. It's purely subjective.

Edited October 20, 2010 by Harry Dolphin

[+bflentje]

Seems like the dumbest idea I've seen in quite a while.

 

Where's MtnMan?

[+Nicodemus3]

Wouldn't it be simpler to just give one point for each 'star'?

 

1 star => 1

2 star => 2

3 star => 3

4 star => 4

5 star => 5

This was my first thought. Caches are already weighted accordingly. I don't understand why you'd want to bring in a logarithmic system over a simple linear model.

[+bflentje]

A few years back I proposed the Star Struck stat.

 

It is nothing more than ALL of your stars summed up and then divided by find count. Seems to be the best way to do what your describing without adding objective weighting.

[+Isonzo Karst]

All the users would need to understand is that 3 stars corresponds to an average "1 find", 1 or 2 stars give you a 1/4 or a 1/2 of "a find", 4 stars and 5 stars five you 2 and 4 "finds" respectively.

 

I'd say the "average" find would be the cache D/T rating found the most - which I'd bet is 1.5/1.5 - it's not the near the middle of the Clayjar scale.

 

A monkey wrench in this notion, whether logarithmic or linear, is the issue of terrain 5. Which often indicates "special equipment" - it's not necessarily the most demanding terrain.

 

I joined a friend on a terrain 5 cache today. A paddle cache, requiring canoe or kayak.

The toughest thing I did all day was get the 'yak down off the car and into the water. After that I sat in the boat, paddled gently, and enjoyed the lovely day on the river.

 

Recently, I found a cache with a 4.5 terrain rating. I "cheated" it, by ignoring the hiking route and paddling. From the river bank, less than 1000 feet of moderately swampy terrain to an easy find. By the standard rating, I turned a 4.5 into a 5. In reality, I turned a 4.5 into a 2.5 - 3

[jd-mitchell]

Not to throw a monkey wrench into the equation, but would experience level figure into it?

[Motorcycle_Mama]

Seems like the dumbest idea I've seen in quite a while.

 

Everyone is reminded to keep all the posts in line with the Forum Guidelines.

 

3. Personal attacks and inflammatory behavior will not be tolerated. If you want to praise or criticize, give examples as to why it is good or bad. General attacks on aperson or idea will not be tolerated.

[knowschad]

Are you referring to the difficulty ratings, the terrain ratings, or both?

[+M 5]

Seems like the dumbest idea I've seen in quite a while.

 

This from a guy with all those mind blowing stats on his profile page, like bearing and distance by bearing etc. It is just another stat that many stat people would like. Like I said earilier, it's more of a stat for you and your friends that would be displayed on your profile page and not after your name.

[+dfx]

I'd say the "average" find would be the cache D/T rating found the most - which I'd bet is 1.5/1.5 - it's not the near the middle of the Clayjar scale.

but that doesn't mean the rating scale isn't logarithmic, it just means that there's a lot of those caches around.

 

i do find the ratings to be roughly logarithmic. the difference between 1.0 and 1.5 is minimal. the difference between 1.5 and 2.0 is slightly larger, but still small. the difference between 4.0 and 4.5 is pretty big. the difference between 4.5 and 5.0 is immense. so that's logarithmic. of course that's my personal perception and everyone is free to disagree.

 

A monkey wrench in this notion, whether logarithmic or linear, is the issue of terrain 5. Which often indicates "special equipment" - it's not necessarily the most demanding terrain.

 

I joined a friend on a terrain 5 cache today. A paddle cache, requiring canoe or kayak.

The toughest thing I did all day was get the 'yak down off the car and into the water. After that I sat in the boat, paddled gently, and enjoyed the lovely day on the river.

but, without your friend, you wouldn't have been able to do it at all, right?

[Hedgehog in the Fog]

i do find the ratings to be roughly logarithmic. the difference between 1.0 and 1.5 is minimal. the difference between 1.5 and 2.0 is slightly larger, but still small. the difference between 4.0 and 4.5 is pretty big. the difference between 4.5 and 5.0 is immense. so that's logarithmic. of course that's my personal perception and everyone is free to disagree.

 

That is why I wondered what would be a good way to come up with a weighted count.

 

I really appreciate all the constructive (and even less than constructive :-) replies above. Let me re-iterate: I take the number with a grain of salt and I advice everyone the same. They are clearly not for comparing different people. It is just there must be a better measure of your own count of geocaching experience however approximate it may be. People talk about the total count all the time. I don't think anything we do would improve the accuracy drastically but some form of weighting may help (or not... I realize the challenge here).

 

Forgetting about the formulas, power laws, etc, if we were to establish a VERY approximate comparison of 1-5 terrain and difficulty rating in units of time or effort, how do you think it would look? For instance:

 

1 star = 1 (minutes, calories expended, whatever)

2 star = 5

3 star = 20

4 star = 50

5 star = 100

 

I don't propose the above numbers, it's just an example to pose the question. Statistics is a funny thing. Even though the ratings of caches may vary drastically from one owner to another or from one area to another, I would expect a nice normal distribution of effort for each difficulty level across many caches just like it happens anytime multiple independent variables are involved. And a normal distribution has to be centered around some mean value - that is what I am after.

 

Unfortunately, rather than a pole, there is no way to gather these stats, is there?

 

I don't confuse the fun aspects of our hobby with statistics and I don't try to get out there every weekend to increase my finds count (but I do get out there every weekend to have fun). I do find this discussion entertaining and hopefully some others will too. Please feel free to criticize and propose alternatives. This is not a discussion of proposed rule changes, just something that a few of us may be interested in to calculate for themselves and their friends.

 

Cheers and happy caching!