Jump to content

Statistics - bend it anyway you like!


Carbon Hunter

Recommended Posts

Is there is simple mathematical formula to get the most NE cache using GPS points?

And what would the centre of South Africa be?

There are two problems with this question. The first is that NE is different from N, S, E and W in the sense that the latter have fixed references (the poles and the Greenwich line) while NE does not. You will first have to define a reference point, presumably the centroid of South Africa. Where this is is something I would like to know as well. (Jors?)

 

The second problem is with the "most" in most NE. What does it mean for one point to be more NE than another? Farther from the reference point does not necessarily mean more NE.

 

I don't know what the "official" method is to determine such points (for example, near Cape Point, one finds the South-Westernmost point of SA), but there is a simple method which, in my opinion anyway, makes a lot of sense.

 

To find the Northernmost point (or cache), one takes a horizontal line (a great circle parallel to the equator) and move it downwards from above the country - the first point (or cache) to hit this line, is the northernmost. Similarly for the Southernmost. The eastern- and westernmost points are determined using a vertical line (a great circle parallel to the greenwhich-dateline circle).

 

So I propose to calculate the SW, SE, NW and NE points, use circles making a 45 degree angle with the equator (or put differently, tilt the earth through a 45 degree angle, while keeping your lines of longitude and latitude fixed).

 

I believe an approximate formula to determine these four points would be do use the values (N-E) and (N+E). The point with the maximum value for (N+E) would be the north-easternmost points, and the point with the minimum value for (N+E) would be the south-westernmost. Similarly, maximum of (N-E) corresponds to north-westernmost, minimum of (N-E) corresponds to south-easternmost.

 

This formula assumes a locally flat earth (that is, in the SA region), so there's bound to be a bit of an error - I'll work on the problem and see if I can get a better formula (or at least an error bound).

Link to comment

I am thinking along the same lines as besem.

 

If you draw a horizontal line on the most northern cache, and a vertical line on the most western cache, and then the point where those to lines intersect would be the most Northwestern cache point. The cache closest to that point would be the most NorthWestern line.

 

Maybe I will try this with GSAK, and then re-order the caches from this custom point, and see what it comes up with. Not mathematically correct yet. But getting there.

 

Would be interesting to see what the 8 Cardinal caches of each province are ... and maybe create a Final cache for all of the provinces by getting clues from the other 8.

 

Hmm ... Only problem is if another cache gets placed with is more extreme....

Link to comment

So I propose to calculate the SW, SE, NW and NE points, use circles making a 45 degree angle with the equator (or put differently, tilt the earth through a 45 degree angle, while keeping your lines of longitude and latitude fixed).

It sounds easy, until you try to do it. You still need a fixed point around which to tilt the earth. To put it in other words, where would your new southwest pole be?

Link to comment

So I propose to calculate the SW, SE, NW and NE points, use circles making a 45 degree angle with the equator (or put differently, tilt the earth through a 45 degree angle, while keeping your lines of longitude and latitude fixed).

It sounds easy, until you try to do it. You still need a fixed point around which to tilt the earth. To put it in other words, where would your new southwest pole be?

 

I think it would make most sense to tilt it along the Greenwich line, so that the north-west pole lies on the Greenwich line itself (and the SE pole on the date line). Agreed, it's probably not going to be a trivial exercise.

 

However, there are some "flat" projections out there (I speak under correction, but I think UTM is one of them?), in which case the (N+E) and (N-E) formulas will produce the correct results.

Link to comment

This formula assumes a locally flat earth (that is, in the SA region), so there's bound to be a bit of an error - I'll work on the problem and see if I can get a better formula (or at least an error bound).

 

Found what I was looking for. Use a Mercator projection (but NOT UTM) to flatten the South African region, then use the 2D formulas.

 

Let N and E be the latitude and longitude of a cache. Then use the value of E+ln(tan{pi*N/180} + sec{pi*N/180}) to determine the north-easternmost point (maximum) and the south-westernmost point (minimum).

 

Similarly, use the value E-ln(tan{pi*N/180} + sec{pi*N/180}) to determine the south-easternmost point (maximum) and the north-westernmost point (minimum).

 

This formula is not a lot different from (N+E) and (N-E), differing from these values by less than 0.05 degrees in South Africa.

 

Can someone apply these to see if it gives reasonable answers? From looking at the maps, I predict the following:

SW: GC1VRJF

NW: GCVV5W

NE: GC1MTXZ

SE: GC1GB58 (this one's hard to judge by just looking at the map - very interested to see what the result is.)

Link to comment

This formula assumes a locally flat earth (that is, in the SA region), so there's bound to be a bit of an error - I'll work on the problem and see if I can get a better formula (or at least an error bound).

 

Found what I was looking for. Use a Mercator projection (but NOT UTM) to flatten the South African region, then use the 2D formulas.

 

Let N and E be the latitude and longitude of a cache. Then use the value of E+ln(tan{pi*N/180} + sec{pi*N/180}) to determine the north-easternmost point (maximum) and the south-westernmost point (minimum).

 

Similarly, use the value E-ln(tan{pi*N/180} + sec{pi*N/180}) to determine the south-easternmost point (maximum) and the north-westernmost point (minimum).

 

This formula is not a lot different from (N+E) and (N-E), differing from these values by less than 0.05 degrees in South Africa.

 

Can someone apply these to see if it gives reasonable answers? From looking at the maps, I predict the following:

SW: GC1VRJF

NW: GCVV5W

NE: GC1MTXZ

SE: GC1GB58 (this one's hard to judge by just looking at the map - very interested to see what the result is.)

 

Suggestion:

Lets get this working in an Excel spreadsheet, and check the results.Then I can work from there and attempt to put it into a SQL query. If I have some time left this weekend, I will try and this formula in a Spreadsheet and see what the results are.

Link to comment

Hi, I have a query - what are the top 10 caches with the most finds in SA?

1. GCMYYZ Table Top Trove: 269 finds

2. GC9CC4 Signal Hill: 223 finds

3. GC43FA Historical Series – Harbour entrance: 209 finds

4. GC3055 Smuts House, Irene: 208 finds

5. GC114RH Cape Town TB Hotel: 203 finds

6. GCWK3K TF16 180° Sea: 183 finds

7. GCAF35 Sea Point Historical 1: 153 finds

8. GC77E Cape Agulhas: 143 finds

9. GCGA0Z Kloofnek Historical 12P: 137 finds

10. GC13FB Pretoria-East 001: 134 finds

Link to comment

We have a question/query request:

 

Which top 10 active caches have the most trackables that have gone through the cache to date? In other words, we would like to know the caches that have had the most trackable items move through them (history).

 

If possible, could you run this query for the whole of RSA and then run it just for Gauteng. It would be interesting to see this because it would show which caches are very safe for dropping off trackable items.

 

Many thanks

GEO936

Link to comment
We have a question/query request:

 

Which top 10 active caches have the most trackables that have gone through the cache to date? In other words, we would like to know the caches that have had the most trackable items move through them (history).

 

If possible, could you run this query for the whole of RSA and then run it just for Gauteng. It would be interesting to see this because it would show which caches are very safe for dropping off trackable items.

 

Many thanks

GEO936

I don't have that type of history in my database. Maybe Danie has the info for the last 3 or 4 months of the trackables, and could work it out.

Link to comment

I can not think of any automated way to determine this - the only data about trackables contained in pocket queries is which trackable is currently where. The only ways I can think of to get the history is to manually inspect each cache listing, or to build up the history with a series of snapshots over an extended time. I am following the latter route, but I do not have nearly enough data yet.

 

If you follow the thread about the TB and GC statistics, you will notice that I periodically list the caches with the most trackables. You will find that these are usually the caches with "TB hotel" in the title. These TB hotel caches are usually safe and locked and are in my opinion exactly what you are looking for. (In Gauteng the following often feature: Little Netherlands TB Hotel, Ki-Deo, Vaal TB Hotel, Pretoria Hitch Hiker Hotel.)

Link to comment

Thanks guys! We also did not know if it was possible to get the "history" and that is why we posted the request thinking that there was a way to determine that. But alas there is no automated way.

 

We are aware of the TB Hotels - we are the owner of the Little Netherlands one - however, there are a few of our caches which are safe for trackables and have had a lot got through them, e.g. Hp in the Park. So it is not just the TB Hotels that should be looked at ....

 

Thanks for the info. though!

Link to comment

The following graph shows the cumulative number of logs over time:

Logs.jpg

 

The next graph shows the number of logs per month for the last 12 months:

Logspermonth.jpg

Any explanation why April was such a popular month for caching?

 

The current rate is 136 logs per day.

(All of these statistics are for ALL logs in South Africa, not just finds.)

Link to comment

Just my thoughts - perhaps April was busy due to the Easter holidays. Christmas holidays many people travel and visit family with the emphasis on family time whereas April the weather is great and less people travel so more time could be spent on social activities such as caching. :D

 

The cumulative logs are growing at a rapid rate! Shows how the popularity is growing. It would be interesting to see the number of cachers superimposed on that graph over the same period of time. I suspect it would be very similar.

 

Thanks for the stats - keep them rolling in. Always interesting to read these logs.

Link to comment

As cincol requested, here is a graph showing the cumulative number of logs and cachers:

Cumulativelogsandcachers.jpg

I plotted it on a logarithmic axis, which means that the number of cachers and the number of logs are both roughly exponential over time.

 

It needs to be said that the number of cachers does not mean all of them are active - I counted every person who logged anything over the total time interval. In a previous message (14 August) I calculated that there are about 524 (now 579) active cachers in South Africa.

Edited by Danie Viljoen
Link to comment

Up to 17 October there are 148 053 logs by 3 918 cachers, for an average of 37.8 logs per cacher.

There have been 111 725 finds by 3 621 cachers over the same period, for an average of 30.9 finds per cacher.

 

This implies an average of 29.5 logs and 22.2 finds per cache.

Edited by Danie Viljoen
Link to comment

The top 10 most active cachers for the year ending 17 October are:

 

iPajero: 1184 finds

gerhardoosMPsa: 490 finds

Danie Viljoen: 481 finds

Henzz: 379 finds

Antron: 378 finds

RedGlobe: 354 finds

DRDM & Raider: 336 finds

cache-fan: 320 finds

Tricky Vicky & Mickey: 314 finds

cownchicken: 312 finds

 

(iPajero has more finds for the year than the next two together - incredible!)

Link to comment

If we don't ALL hide more caches soon, then iPajero will not have anything to do next year.

 

Nonsense - iPajero can then trek around the country again, placing caches as he goes. That'll give us more to find and a better chance to catch up with him!

Link to comment

I did a quick check - there are only about 1400 active caches left in South Africa for iPajero to do. At their current rate it will take them less than 14 months, although they will of course not be able to maintain this rate as the available number gets less and less!

 

But that is only if iPajero do all their caching in RSA....if they decide to go on a little holiday in Europe, they will have tons more caches that they can do! I was talking to Rod of RodnJoan at the recent Event by Crystal Fairy, and he was saying that in the UK, where his daughter lives, there were 500 active caches in a 6km radius!!! WOW. I also know that in Germany there are tons of caches, as the Germans are just cache crazy, so if you land up in Deutschland for two weeks, my guess is you could rake up a good few hundred caches! :laughing:

Link to comment

Yes - most European caches have their descriptions in the local language - i've done caches in France, Netherlands, Belgium, Germany and Spain. Many are bilingual - but the exception.

 

Netherlands is very good at having bilingual caches.

 

Babelfish translation page online does help with the traditional caches - but puzzles and multis can be a bit harder.

Link to comment

I am looking at alternative ways to determine the real difficulty of caches. One such metric is the percentage of DNFs. (100*DNFs/(DNFs+Finds))

This will not work for puzzle caches, which normally do not get logged if the puzzle is too difficult. The following caches are the top 10. (To qualify for this list, a cache has to have at least 5 finds).

 

1. GC1RZCV Table Mountain Cableway: 60% (D=3)

2. GC1JQ4V Cache-a-pult: 54% (D=2)

3. GC1AN54 Big Guns of Ladysmith: 50% (D=3)

4. GC1M3K5 DTI Traditional Gate: 50% (D=3.5)

5. GC1E2XQ Rooikrans Crossing: 50% (D=3.5)

6. GC1FAQC Karkloof Falls Jr.: 50% (D=3)

7. GC1GT9Z Paul Roux: 50% (D=3)

8. GC18KRV Looking Through: 46% (D=2)

9. GC1HNH0 Harry's Masters: 45% (D=2)

10. GC1QVD4 Genes, Geese and Green: 45% (D=1.5)

 

One would expect all of these to have a difficulty of D=5, but as can be seen above, some are as low as 1.5!

Link to comment
1. GC1RZCV Table Mountain Cableway: 60% (D=3)

 

Ahem, Ohh, now I feel a little guilty. That's mine.

Just to qualify, it wasn't meant to be hard. I have a tradition of not giving the hint till the FTF has found it. (This is a nano hidden on a building with a few tons of steel railings on it). I have changed the listing a lot after it was found the first time, including a "if you really can't find it" encrypted hint which spells out where the cache is hidden.

 

Problem with the rating system used is that listings are dynamic and DNFs may reflect history, not current difficulty.

Link to comment

With quite a few archived caches going through lately, I have had to update my PQ to get all the active caches in South Africa to fit into 7 or 8 queries.

 

In case you are interested in getting them all here are the (names I use and ) dates required to get them all in 8 queries:

ZA - All Caches 1 - 1 Jan 2000 to 12 Mar 2006 (498 Caches)

ZA - All Caches 2 - 12 Mar 2006 to 23 Dec 2006 (494 Caches)

ZA - All Caches 3 - 23 Dec 2006 to 15 Sep 2007 (497 Caches)

ZA - All Caches 4 - 15 Sep 2007 to 15 Feb 2008 (492 Caches)

ZA - All Caches 5 - 15 Feb 2008 to 10 Aug 2008 (498 Caches)

ZA - All Caches 6 - 10 Aug 2008 to 5 Mar 2009 (497 Caches)

ZA - All Caches 7 - 5 Mar 2009 to 29 Aug 2009 (497 Caches)

ZA - All Caches 8 - 29 Aug 2009 to 31 Jan 2010

 

How to create the PQ:

Add the above name, or whatever name you want, then in the PQ select only the following:

Show me 500 caches

Any Type

Any Container

Countries: South Africa

Between [Date1] and [Date2]

Edited by DamhuisClan
Link to comment

The following statistics give the percentage of available caches found by the leading cachers for South Africa, for each province, as well as for the rest of Africa:

 

South Africa: available: 3701

iPajero 2253 60.9%

Tricky Vicky & Mickey 1263 34.1%

cache-fan 1102 29.8%

cownchicken 1094 29.6%

RedGlobe 920 24.9%

 

Western Cape: available: 864

cownchicken 752 87.0% (wow!)

battlerat and pussycat 658 76.2%

Tricky Vicky & Mickey 519 60.1%

The Huskies 512 59.3%

iPajero 488 56.5%

 

Gauteng: available: 786

cache-fan 610 77.6%

RedGlobe 568 72.3%

GreenJaM 565 71.9%

Antron 460 58.5%

KeithWood 424 53.9%

 

KZN: available: 645

iPajero 467 72.4%

Kwenda Tafuta 440 68.2%

DiePienaars 393 60.9%

Noddy 389 60.3%

Blackjack Bailey 374 58.0%

 

Eastern Cape: available: 448

iPajero 366 81.7%

iNokia 332 74.1%

cownchicken 191 42.6%

ravenmeistre 172 38.4%

amaSoekSoek 158 35.3%

 

Mpumalanga: available: 393

gerhardoosMPsa 254 64.6%

iPajero 221 56.2%

Fish Eagle 219 55.7%

cache-fan 185 47.1%

Bouts777 176 44.8%

 

Limpopo: available: 218

Mapula 111 50.9%

iPajero 111 50.9%

Fish Eagle 109 50.0%

henslin 107 49.1%

3sofar 77 35.3%

 

Free State: available: 147

iPajero 119 81.0%

Tricky Vicky & Mickey 64 43.5%

gerhardoosMPsa 63 42.9%

veer-ini-hoed 56 38.1%

Urban Hunters 56 38.1%

 

North West: available: 142

RedGlobe 94 66.2%

iPajero 89 62.7%

cache-fan 76 53.5%

Fish Eagle 70 49.3%

Antron 53 37.3%

 

Northern Cape: available: 58

cownchicken 30 51.7%

iPajero 26 44.8%

The Huskies 19 32.8%

Tricky Vicky & Mickey 19 32.8%

Bannavdm 16 27.6%

 

Rest of Africa: available: 625

cownchicken 65 10.4%

Tricky Vicky & Mickey 62 9.9%

Udjat 59 9.4%

kokoscoot 50 8.0%

Landy-Team 41 6.6%

 

iPajero leads in 4 of the 11 areas, and cownchicken in 3.

Link to comment

If one takes into account the fact that iPajero owns 70 active caches in the Eastern Cape, then they have actually found 96.8% of the available caches in that province!

 

For the same reason the real percentage for cownchicken is 89.5% in the Western Cape.

 

Incredible.

Edited by Danie Viljoen
Link to comment

If one takes into account the fact that iPajero owns 70 active caches in the Eastern Cape, then they have actually found 96.8% of the available caches in that province!

 

For the same reason the real percentage for cownchicken is 89.5% in the Western Cape.

 

Incredible.

Love the stats - that is amazing CnC & iP - great stats for those provinces.

 

Rest of Africa also looks very interesting.

Link to comment

With quite a few archived caches going through lately, I have had to update my PQ to get all the active caches in South Africa to fit into 7 or 8 queries.

 

In case you are interested in getting them all here are the (names I use and ) dates required to get them all in 8 queries:

ZA - All Caches 1 - 1 Jan 2000 to 12 Mar 2006 (498 Caches)

ZA - All Caches 2 - 12 Mar 2006 to 23 Dec 2006 (494 Caches)

ZA - All Caches 3 - 23 Dec 2006 to 15 Sep 2007 (497 Caches)

ZA - All Caches 4 - 15 Sep 2007 to 15 Feb 2008 (492 Caches)

ZA - All Caches 5 - 15 Feb 2008 to 10 Aug 2008 (498 Caches)

ZA - All Caches 6 - 10 Aug 2008 to 5 Mar 2009 (497 Caches)

ZA - All Caches 7 - 5 Mar 2009 to 29 Aug 2009 (497 Caches)

ZA - All Caches 8 - 29 Aug 2009 to 31 Jan 2010

 

How to create the PQ:

Add the above name, or whatever name you want, then in the PQ select only the following:

Show me 500 caches

Any Type

Any Container

Countries: South Africa

Between [Date1] and [Date2]

 

Anton, just a note... I do mine very similar but rather than doing the same start as the previous end date I make it the day after ie: where you have 12 Mar 2006 to 23 Dec 2006 I have 13 Mar 2006 to 24 Dec 2006. This method has given me and extra day to work with and later on it gave me an extra week. Of course this may mess us all around when some archived caches become unarchived. But it just requires us to take note of sudden changes.

Link to comment

... I do mine very similar but rather than doing the same start as the previous end date I make it the day after...

 

Agreed, overlapping dates aren't neccesary.

Also, there is handy Gsak macro to keep things optimised, calculating the dates for you, once a maximum number of caches are set.

Setting the macro to calculate at a max of 498 caches per query works well, leaving some room for unarchiving.

Link to comment

Tried it and I still get a better result by some trial and error. Most of these were nice and rounded to the high 490's just last week... But this week some more must have gone under the Fish Eagles mighty claw...

 

My list of PQ's are as such

 

Waypoints in each file

 

(PQ1: 01 Jan 2000 - 12 March 2006) 498

(PQ2: 13 March 2006 - 24 Dec 2006) 496

(PQ3: 25 Dec 2006 - 15 Sep 2007) 491

(PQ4: 16 Sep 2007 - 16 Feb 2008) 489

(PQ5: 17 Feb 2008 - 19 Aug 2008) 494

(PQ6: 20 Aug 2008 - 12 Mar 2009) 495

(PQ7: 13 Mar 2009 - 3 Sep 2009) 495

(PQ8: 4 Sep 2009 - 31 Dec 2009) 239

 

I like to keep it in at around 390 - 395 should there be any caches that get unarchived. I would be able to play round with PQ 3 and PQ 4 here and add a day onto both. Depending on how this plays out I may have some play on the rest. But then again a few caches may get unarchived and I have to go back... So I will give it some time and see what happens. Or maybe not.

 

The macro run on my Database generates the results below. I can see that in conjunction with some manual playing round it will help a little in getting a final PQ setup. Thanks Jors for mentioning this and Beejay & esskay for the link.

 

Placed by date PQ Generation

PQ From To Count Done

01 2001/01/18 2006/02/27 489

02 2006/02/28 2006/11/15 490

03 2006/11/16 2007/08/04 486

04 2007/08/05 2008/01/06 485

05 2008/01/07 2008/06/20 490

06 2008/06/21 2009/01/01 490

07 2009/01/02 2009/07/02 487

08 2009/07/03 Maximum date 461

 

Just wondering does anyone do it differently, maybe doing a PQ or two per province or cache type? It may be a bit ambitious for some to run a database for the entire country seeing that the chances are you probably won't be as well traveled as iPajero... I just run it like this cause it is great to have every cache at hand should I want to see them offline.

Edited by Wazat
Link to comment

Agreed, that the overlapping dates are not required, but I am trying to error on the side of caution here.

If some caches should be unarchived, and pushes the count over 500, then the next PQ should catch the missing caches of the last day of the previous PQ.

 

(This is working on the assumption, that the caches are added sequentially to the PQ per date)

 

Luckily GSAK highlights in red when a PQ is 500. So the error would be caught quite quickly.

 

As a side note:

I used to save all PQs to a directory, from my email client, and then drag and drop each ZIP file onto GSAK, and import all that way. I found out the "Get data from email" allows one to read from email and from a directory. Now I import all 8 zip files in one go.

Link to comment

All the caches in South Africa have been planted by a total of 730 persons, which gives an average of 5 caches per planter.

 

The most prolific cache planters in South Africa are:

CrystalFairy: 161 (100)

Fish Eagle: 133 (86)

iNokia: 76 (73)

Noddy: 75 (62)

iPajero: 73 (73)

 

(The first number is the total number of caches planted; the number in brackets is the number of caches still active)

 

The top planters of large caches:

Fish Eagle: 8

batsgonemad and his squirrel: 2

gerhardoosMPsa: 2

radebuddyz: 2

 

The top planters of regular and small caches:

CrystalFairy: 95

Fish Eagle: 94

iNokia: 55

Noddy: 46

iPajero: 45

 

The top planters of micro caches:

CrystalFairy: 55

NotBlonde: 36

hennieventer: 26

Gps Storm: 24

cache-fan: 23

 

The top planters of multi-caches:

CrystalFairy: 14

Discombob :12

Noddy: 12

Fish Eagle: 11

CapeDoc: 10

 

Lastly, the top planters of mystery caches:

CrystalFairy: 30

Cism: 27

Donkasand: 10

Rhino and Hedgehog: 10

River God: 10

Link to comment

Danie/Anton,

Maybe this is my imagination. I have a bookmark on some of the caches and I also receive the logs on my own caches. During the last two months I noticed a couple of new cachers and it feels like that they are on the increase.

 

Is there any way you can measure the growth of cachers in a form of a graph? Maybe use the last three years on a monthly basis to answer this one. Sorry for the homework.

 

Gerhard

Link to comment
Danie/Anton,

Maybe this is my imagination. I have a bookmark on some of the caches and I also receive the logs on my own caches. During the last two months I noticed a couple of new cachers and it feels like that they are on the increase.

 

Is there any way you can measure the growth of cachers in a form of a graph? Maybe use the last three years on a monthly basis to answer this one. Sorry for the homework.

 

Gerhard

 

This could turn out to be an interesting one.

I will try and create a query to see how many people made their first find in each month.

I think that would answer your question.

Link to comment
Guest
This topic is now closed to further replies.
×
×
  • Create New...