Is the average CR4 member in Timbuktu ?

recommending my friends in India to be members and acquire sea of technical knowledge from this Forum.

Good for you !

For those who haven't noticed, hover your mouse over "Tell a Friend about CR4" at the top of the page.

Is that an oxymoron?

I suppose 'crouch your mouse' might work ? ah....friends/geeks/reality.......

...........'crouch your mouse' might work ?

How about Xela Hcuorc..........his personal power animal is a spotted skunk!!!

I am not quite sure whether this bears the remotest resemblance on what you said.........but, I felt that I had to say something!!!!!

I think that I have finished things today at 0115........... I have to get up for work this morning............sayonara (or something like that)

onara

you did tell us to say it, didn't you?

I've been waiting ages to find out who 'Us' is. Now all I need is the find out who 'They' is. It's about time I went off to multiply, and give some kind of answer to the problem I proposed

you don't multiply

We do .

For us us is Us and they are you.

For you us is USA or USSA or US they are.....( i don't want to be political or CL will close the thread) .

You misread that sb................he said..............go forth and multiply...

You are probably having trouble with your computer again.............it must be the quality of the electricity.....

..........the poor electrons................they go hither...............the go thither.................... they know not which way to go.............no wonder you have problems sb.

Kris, In relation to original data, could you give basis for determining X & Y for Northing and Easting. Also, by using 0 Longitude for UK eliminates X-weighting for UK members - is this appropriate? It could put us on the coast instead of the desert in the final result.

Sure,

I just read the coordinates for each city off Google Earth. Americans are all in Troy (just for fun), and Australia got inverted in a typo. Figures are rounded up.

Although it may not be very clear in the image, locations have been converted to xyz co-ordinates (Centre of earth is Zero). This is to avoid antipodal points averaging out to zero, and also eliminates any position having x,y, or z as Zero. Radius is taken as 1 to simplify calculations. The final result can be converted back into latitude and longitude. As Gwen pointed out earlier, the derived 'average' position inevitably falls within the Earth. It was necessary to do that in order to provide a relatively simple model (centre of gravity of data points). Extending the c.o.g. through to the surface works well enough at this stage.

Nobody has yet contested the fundamental approach, so I shall shortly be re-working it by way of angular separation. I'll explain the why/wherefor of that at the time. If it's of huge interest to people, I can re-assign the location within each country. I'm a bit reluctant to do so, since there could be a lot of debate about what location to use for each country. In any event, the first part of my 'answer' will ignore membership of each country. It will hopefully make things clearer if I do that first, then superimpose weighting according to country membership. The diagrams will explain this a bit better than words alone can.

Hope that helps a little.

Troy, N.Y.??? Are you sure? Have you seen Troy? Nice Wal-Mart!

A lot of pumpkins got murdered there last year. Honest - they even showed the pictures, quite brazenly, on CR4.

Am I the only one who is thinking, wait, lets say you have 2 CR4 users in Oregon, and 2 users in New York, that means that the average CR4 user is from Iowa? Even thought when you look at your data set, you know none of the users live in Iowa, or anywhere near there...

It's no wonder so many people don't like math! 8^) (Hey, I have a BS in math, so I can make fun of it)

You know: figures don't lie but lier's do figure.

Stay tuned - you're gonna love thrashing my 'answer' !

Kris,

I can barely read you're spreadsheet. Would you post it a bit larger so I can read it, either in a response to this or an cr4-mail to me?

It's funny how psyched I am that you did this. Really, great job. Sure it's a waste of time, but what a fun waste of time!!

Roger

Hi Roger,

Glad to see I've lured you in

Yeah, the spreadshit was a pain to squash in. I'm running a bit late, but will see if I can re-post it in a clearer from tomorrow. Meantime, you should be able to enlarge it a bit by pressing Ctrl + "+" or Ctrl + scrollwheel. Watch out for the slight boob with Australia - last time I heard, they were in the South !

As I say, I'll get something sorted within the next 24 hours.

Just to be fair and let everyone know;

I've mailed Roger the list of Northing's and Easting's. I'll paste it up here on request*, though it will appear fully on Thursday. For the present I'm guessing that most people will just be pondering the method, rather than wanting to crunch the numbers. As mentioned earlier, there is a typo (the sign) in the locations for Australia and Panama. That will be rectified. I won't be adjusting the number of members/countries as originally posted, since it has little significance to the end result (though I can easily insert the data if anyone desperately wants me too).

*apply below, or by PM.

If you don't keep abreast of the posts, I can't help you ? _{I know nothing of alien body shapes, much less what position they like to do things in.}

Yep, them fishes/dolphins are just going to have to lump roe it. . If anyone signs up for a very remote Island (it must have country status, just to be fair), I'll consider including it in further analysis. Deciding where it belongs is the essence of the problem. It would be divided, but exactly how is the question.

70% excluded is not a problem. Having both hemispheres included is what matters - it would be less exciting with only one.

The distribution starts off with dots, 36 to be more precise. For added difficulty/confusion, I added country membership. If you ignore the country membership, and just look at the locations for a moment ; You could find one point that is the average, perhaps akin to finding the point that is nearest to all locations (?). It may be that 2 focal points would be a better descriptive. At the absurd extreme, you'd have 36 focal points, but that would be doing nothing with the data (!). The aim is to describe the average location in briefest, but most accurate manner.

I could have made this easier - people positioned in a field. Where do you stand so that you need the smallest length of string in string-telephones to reach all. I didn't, so you can ignore this simplification.

Since you are being nicely curious, I shall explain more ; In geology, discontinuities often exist in particular orientations. The planes have a direction of dip, and a dip angle. It is quite valid to describe such a plane by defining the normal (ie perpendicular) line to it. Imagine all these lines drawn thru the centre of a sphere. Where the lines intercept the sphere will depend upon the lines orientation. Just one half of the sphere need be considered (since the other half contains the same information, just in antipodal form). This hemisphere,with all it's dots, can be projected onto a 2D map - a circle. Methods exist to interpret the data so plotted. With rock, it is quite common to find 1,2,3 particular directions that planes tend to lie in (the trend is some times referred to as 'hade'). Geology is not always kind, and the picture can be a bit blurred. All the same, you have to make a best-guess. Methods exist, but are open to debate. Note here, that points diametrically opposite on a circular plot are almost the same in orientation. Not a problem - it can be dealt with.

Above, you have the images of such plotting with rock. Although the data represents 3D structure, it's not the same situation as the question as posed. You can't put two hemispheres on the same circular plot (for the Earth's surface locations, antipodal points must retain their individuality) Using 2 circular plots (one for each hemisphere) will not help you. The stuff above is very 'crude' in terms of current methods. A vague 3D analogy could be made. Manipulating angular separation mathematically is better methinks.

To analyse the problem posed, you need to stay within purely numerical analysis.

I have no idea who might consider the problem - astronomers, meteorologists, whale-watches. No idea, but applications probably exist. All that's needed is to see beyond the "Is the average CR4 member in Timbuktu ?" smokescreen. People who don't recognise the problem might have fresh insight that others don't know of. Even those who look at star maps etc, might be lured into thinking afresh about how they decide a particular area of the sky has a 'cluster'.

The string telephone analogy still works (if you can imagine one that works around curves) - where do you stand/float so as to speak to all individual members, having spent the least amount on string ? Radial connections only. Not a good analogy, because 2 or possibly 3 clusters might be better for the given data set. The number of American members seriously skews the data, but even determining the offset from Troy,NY would be a start. Needless to say, I have my own answer to this already. When so inclined, I'll look at the best fit for 2 clusters - it would be much more illustrative of where all members in the data are located. As mentioned, I consider it easier to do this by computer program. It could churn out images of the globe (viewed in pairs of 2 pictures from opposite sides). There are many methods of tweaking the way in which cluster location is found.

It's maybe best described as a 'think outside the box sphere' problem. Method, and justifying it, is more important than crunching the numbers. The numbers are simply available for those who wish to play, or be able to criticize my interpretation. If people prefer the sand pit to the roundabout, that's fine.

Solutions are often simpler than problems. Given the task of finding the volume of an object one can: look up a known formula, invent a specific formula or think of immersing the object in a water filled graduate.

The simple example given by Kris, minimising the length of a string connecting positions is not so simple in real life: Are the positions on a plane? On the surface of a sphere? In 3D space? If they are on a sphere, are we allowed to dig tunnels? If all points communicate with a switchboard, are 'swithboard busy' signals allowed? How often? Can multiple switchboards be used? What is the relative cost of wire, of digging tunnels and of switchboards?

Or, back to the example of CR4 centroid of members, should each of the 36 points be adjusted for population size? For the proportion of English speakers? For the size of available engineering colleges? For the ease of using the internet? Should we compensate abscence by creating virtual members?

In this day and age, when canned solutions are easily obtainable for most problems, it may be more important to expend effort in defining aims than in importing solutions from seemingly distant disciplines (which has often been rather well done by MatLab, SPSS and others anyway).

Thinking out of the box is highly important certainly, but should, I think, be better used in defining aims and translating aims into problems. Kris is to be praised for goading so many members into illuminating and emotional response.

dovy,

Many thanks for your comments, and the complement.

Canned solutions are indeed a problem. At many levels, people chuck numbers into equations without having any idea of how the equation works. This extends to software solutions. People need to understand the basis of the tools they use. Stretching the brain is never a waste of time. I'm sure many people have been caught out by off-the-shelf software solutions that were faulty or not appropriate.

There's another analogy I could use from the CR4 data. Astronomers will probably scream at this, and say that they already tackle such problems. I know they do, but I wonder how many have sat down and spent time evaluating the validity of their model. Here goes ;

You have a telescope that can locate stars, and they are all the same distance from Earth. You want to optimize where you look for ET. The Earths longitude and latitudes have been extended onto a celestial sphere, with a grid separation of 1^o. Within each 1^o x 1^o sector, the number of stars ( = CR4 members) is known. The task is to decide which sector a rocket should be sent to. Once in position the rocket cannot move, and it can only call ET (who lives on only one star) along the surface of the celestial sphere. Where should the rocket be sent in order to minimize telephone time, and can an argument be made for sending 2 rockets to search from different areas.

Like all analogies, it breaks down if over-thought, but the principle is the same as the problem posed. Where is the cluster point(s) on a sphere. Maybe people can see a similarity to problems they actually tackle, and maybe they can't. Whichever, it requires more thought to come up with a methodology for this than many of the Challenge Questions. It's just a simply posed question, phrased so anyone can consider it, and those with related knowledge can gain some cross-pollination of ideas. Interesting that so few have considered it, given that CR4 is a discussion place for engineers and scientists. They must be busy, obsessing about politics and religion.

Cheers for now, if I don't join Facebook I'll be back.

Kris

Kris,

Still, going along now with your search-for-ET analogy, the framework can become richer and more interesting if you define it more realistically. If your spaceship can carry only one communication device, what range of frequencies should judiciously be chosen - based on stray frequencies captured from various directions. Without such considerations it all degenerates into an exercise in 3D projective geometry peppered with some statistics.

Also, not all uses of readymade-solution-makers are bad. After all, most cooperative engineering ventures depend on some sort of belief in a cooperating expert's expertise without being able to check it personally.

And, what is 'facebook'?

dovy

There's nothing wrong with ready-made solutions as such. Not much would get done without them. It's just wise to take a look at the results sometimes.

I was joking when I said I'd go to Facebook. Sticking pins in my eyes would be preferable. It's a sort of 'lets make friends' site. I honestly cannot put into words how vacuous it is.

"I shall run and hide whilst the onslaught begins................"

OK,

Now can you make it stop?

Being too lazy to do it myself, I actually enjoyed the raw numbers.

The rest of it gives me a headache.

Thanks,

LL

It'll probably get a lot worse when I post the more detailed analysis. Because of the limits on picture size, I'll illustrate each step (probably using one row as an example). The aggregate results will be typed in for ease of reading. I've also set up a computer program to do the same as the spreadsheet, but in more detail ().

The quality and quantity of emotional, irrelevant and unfunny response to this somewhat unconventionally presented project are surprising.

Could it be because of members' sensitivity about their geographical origins?

Could it be because of the imperfect way statistics is taught?

Distributions on a circle are a difficult subject both theoretically and practically, not only because averaging 1 and 359 yieldss 180. Perhaps the main difficulty is that both tails of a probability density function coincide. Various ways of treating circular distributions exit, none of them is trivial.

Investigations of population densities on a 2D space are intuitive and familiar. The scale factor: meters, miles, towns, neighbourhoods, counties, countries, continents, planets is neither intuitive nor trivial. Densities are usually assigned to a NAME, regardless of inhomogeniety WITHIN a name.

One practical and well known use of a multidimensional distribution having a circular component occurs in Dispersive Meteorology where the dimentions are: wind speed, horizontal wind direction and vertical wind direction. This discipline serves in predicting airborne pollution dispersion.

I have no idea what Kris is driving at. If he chose an unconventional presentation of the problem, a CR4 member could be entertained instead of using the occasion for letting off repressed steam.

Thanks, dovy.

You expressed that far nicer than I did to a certain person last night. The real 'meat' of the question will emerge later this week. The overwhelming majority of post are fun to read, and I'm pleased that my slightly off-the-wall method of raising a topic has generated so much interest. When my full explanation emerges, hopefully people can gain an understanding of something new. It will be possible to still have fun using it in relation to the topic as posted, and some members might enjoy it as an additional tool in whatever field of work they're involved in.

The manner in which I phrased the question (and my somewhat frivolous track record) was bound to lead to a fair amount of hilarity and fun. That's great, because I want people to look at the answer I intend to post from a wide variety of perspectives. It's all about seeing how people do things on the other side of the fence, and hopefully learning more from each others methods. The occasional bit of invective is no big deal. All the rest of it is what makes CR4 work so well. Before the week is out, we can all emerge with a few laughs and a little more knowledge.

Many thanks for the excellent post.

Kris

Damn your Squirrelly hide, but I actually started thinking about this stuff this morning.
In terms of analysis, maybe the elecrtical concept of Root Mean Square could be useful for getting rid of nasty negative values, and I'm sure if you choose your origin wisely you could get the mean to be anywhere you wanted.
Del

ROFL ! Have another cuppa

The origin used in the spreadsheet is zero. If you ignore the typo on Australia, you should be able to see the direction of the axis. If you were looking at a globe with Blighty to the front, East (X) is positive, North (Y) is positive, and the other one (Z) is poking you in the eye (unless you turned around to pick up your toast ).

As dovy noted, I used xyz because Polar bears and aliens radians scares a lot of people. Almost anyone can understand xy on a normal graph, so it makes it easier to work it up to 3D with xyz. From there, it's only a small hop to get angular separations.

If you start using you electrickery i, I'll cry ! Could be fun......hold your horses cats mice for a couple of days and I shall reveal all. Your rms ac might match the power of my dc directitude. I'll rectify you if it's better ! Here's a very oblique clue....

Your rms ac might match the power of my dc directitude. I'll rectify you if it's better !

..........and is that with a power factor of 1 with a purely resistive load???

I suppository you're right !

.....that's a very anul comment!!!

I'm also still thinking on the best way to do this.

When averaging points in a Cartesian 3D system you end up near the middle of the globe, extrapolating this location to a point again on the surface is not so bad from the idea but you end up in a location where nearly nobody is really interested in CR4 and what it means to the world, they have other problems to solve first. (We could help them if they wanted technical help.)

Leaving the polar coordinate system as it is and averaging then someone in Ireland with someone in Belgium brings us to the other side of the globe, again not what we wanted and certainly wrong.

We can't use the mean here as it is non normal data (somehow you went from continuous to discrete and back to continuous)

What about working out the proportions of CR4 members to the total country population and look for the highest?

Or just look to the sum of GA's per country as these are representative for what we want to achieve: give good answers where another engineer can build upon. (again a mode approach)

This tread looked a stupid time consuming silly thing to do, but deep inside there is a nice problem to solve.

Gwen, you need to look more at the angular separation of points on a sphere. There are 2 screw-ups in the original table locations (Australia and Panama), but it's no big deal within the overall aim of the thread. I shall illustrate how it can be done, and why, in a couple of days. The use of country names is just to make referring to data points easier, and helps provoke a bit of interest in what might otherwise seem a dry topic.

The interesting part is what you do with the problem from there. View the member totals as a group of dots at the same location. France, for instance, would equate to 2 dots at 49^o North, 2^o East. The game is to provide a meaningful description/summary of where all the dots are on a sphere. At one extreme you could say that everyone is located in America (troy to be more exact). At the other extreme you could say we are all smeared across the world evenly. One stage better might be to say there are 2 main regions where members are concentrated, America and India. Exactly how you might quantify/illustrate that situation is a more tricky proposition.....Do I allocate Turkey to the Indian group, or to the American group ? hmmm......possibly split them.....there might not even be 2 cluster around America and India....

http://www.squirrelstop.co.uk/?gclid=CJ_hotWbu5kCFQ0gQgod83Ji5A

Only a bit of eavesdropping . I just don't know when to stop.....

Kris,I think I might have a more targeted approach. First, look at this list given in post #7:

USA 128

India 64

Canada 30

UK 20

Australia 19

New Zealand 11

south Africa 9

Belgium 6

Egypt 6

Pakistan 5

Philippines 5

Notice something? This list is dominated by former members of the British Empire. Because the Philippines used to be a US possession, you could count it as within the sphere of British influence. The same would apply to Egypt, for it used to be a British protectorate.

This not to say that CR4 members from outside of these areas have no value or importance to CR4. The British Empire represents where the largest portion of CR4 come from.

My approach would be to draw a outline around the UK, extend to the US and Canada, go down to Australia & New Zealand, then up to the Philippines, over to India and Pakistan, on to Egypt and end at South Africa. Plot the extreme points of this area, draw lines connecting the opposite corners and see where they cross.

Of course, you will probably continue doing what you're doing (most skwirls do). But you did ask what other people would do. And if this is what I would do if I was going to do this.

I may nominate you for an OBE ! Perhaps I should have dusted off an old atlas, where the world is mostly pink (that phrase would confuse todays youngsters !).

Having invented the internet, it's only fair that everything should be centred on Britland. _{<c'mon America, I know you hate to admit that} > Hey, maybe I'm not so off the wall here.....

or here...

oooh.....

I think you promised some results?.....

.....I live in a slightly different time-frame to the rest of the world ! All in due course.....

Is that Squirrel time or squirrelly time?

well you know anyone that has to calculate normal.... isn't!

...it's possibly hereditary - being of (I'm guessing !) Angle descent, and tending to go off at a tangent . Besides, eager people want my drawing/model making skills on the wave thread . If I spread myself any thinner, I'll look like a bear-skin rug

Kris,

Is it likely with the passing of time that the original data is no longer valid? The membership base has possibly changed with a consequent shift of the centroid and there may also have been some geographical shift in the land masses which will have an affect on the accuracy of the result.

Indeed it is, Bob.

I can plug in the membership changes OK, but I'm not sure how often they correct Google Earth for continental drift.

Can I ask that concerned members slam on the breaks when driving in the appropriate direction - this will help to delay the effect until I produce results. In the interest of expediency, members will have to check upon movement direction and rate for their location. No compensation can be paid for excessive tyre wear. I'd also like anyone in Scotland to move south, in order to compensate for glacial rebound which might mess up the angles.

i will try to counter move when there is an earthquake here

Excellent - and AussieBob must control the roo's.

All under control here

LMAO any calculation that is so slow that it must account for continental drift is news to me. This would be any of that new math they been teaching in the U.S.?

perhaps some bees can help figure out this problem?

hmmmm......

.....you can sort out the shortest connecting path between points with bubbles: Two plates (perspex), separated by short rods (each representing location of town or whatever). Dipped in soap solution, a film of soap will cling between all points. The surface tension will minimize, leaving it to intersect at optimum connecting points. Not a brilliant description, but I recall seeing it demonstrated a long time back. The connecting points for the route will also form 120° between groups of three points.

Some traffic/network expert will pop up to demolish the above, but what they hey. Don't care because I'm now trying to recall the name of the classic problem about crossing the bridges in some German (Dutch?) town. Mwahhhhhh. Back leter after some googling. Might even have to pull a book off the shelf - best get me dust-mask first

Cheers for the link, Chris. It's on my cogitation list. Also on the list is something vaguely recalled about a type of butterfly. Migrates up North America, lays eggs, then dies. The new generation somehow know how to get back South where parents came from. I might have to pause to jelly some confounded eels, or maybe grill a salmon.

The bees were on the news this morning. Buzzing about. Irritated by the researchers' artifical flowers I think!

The soap bubbles were used to plan the UK's major motorway (autobahn/freeway) network. The old black'n'white film used to be shown in schools...

Monarch butterflies.

Confus-a-cat, a subsidiary of Bewilderbeast.

Well hey, if a bee can do it, surely a squirrel can...

What is the shortest distance between two nuts?

A telephone

How did this ever pan... err, sphere out?

With me still looking at it.

The intent was to see how engineers from a different background would tackle this. I know full well how conventional thinking goes. In order to disgiise the situation, it was presented as is.

A few members were dismissive, and a few others very helpful.

I used the worldwide distribution of CR4 members as as a sample, intent being to analyse geodata in a way that was understood by all. The 'weighting' of America made this less than ideal.

The essential problem is averaging data points on opposite sides of the globe. The problem involves 'set probability'. A ponsy way of saying 'weighting'. I know how it's done in conventional ways, but thought it might be interesting to see other;stake on it.

I got bored, despite having worked up several software solutions. Send me a message if interested. As with others who have helped, I've said I'll keep them informed but not the knuckleheads. You are welcome to PM me

The 'weighting' of America...

what are you really saying?

Yer all fat bas******'s !

ROFL - I got the maths done, and even a simple routine to rotate it for various views. Its been a long time, but I sure recall your offers of help. The 'cat is out of the bag' so to say, but I still think it a fun diversion. Calculating primary stress is no easy thing. Once done, it's no big problem to rotate those values onto another axis. Purely numerical, and I've done that <hold yer breath> using BASIC. Stop laughing, it works ! It ain't perty, but it works.

My interest has long since wained. The program in BASIC I had was fun - rotate the cube and colours changes plus the matrix of values changed. The math and physics perfect, just not good looking.

I've not bothered checking what's on the market, and don't have any interest

I'm probably jumping the hoop, Chris. Geotechnics is not usual CR4 stuff. Having been a good mate on this issue, I'll PM you info on it first.You've been a good mate on this one. It was only ever a curiosity for me, but you may have the graphic design skills to make some $$$. I have an idea - yours to use if you want. Long time since I looked at this, and I know you do stuff just for 'fun'. We'll talk soon

Yeah... I am still interested. I was deployed to Iraq when you originally posted it in 2009 and didn't hop on the train until late 2010.

It proved a needed distraction as I tried to readjust to non-war-zone life.

I ran your data manually using my circle method (post 257) but 2D (post 260)... for the record... never doing that again! But it was fun having a reason to break out a long-in-storage manual drafting set.

The biggest decision was where to split the map and based on distribution, I chose the Pacific which was a good thing because I could not find a single map (as shown in post 260) split through the Atlantic and having the Pacific in the middle.

So, my results... garnered through manual calculations, long-hand, and manual drafting (remember, there was a therapeutic side to my actions akin to Buddhist Sand Art) we all ended up in the Atlantic, just west of the mid-atlantic ridge near the Sohm Plain... hope you brought your swimsuit, haha.

Anyway, I'd be interested in what you came up with. I'll send you my contact info in a PM.

************* Magnificent to hear you are back, especially safe and sound !

I'd been getting a bit a bit bored of CR4 (not that I could compare that with what you've been up to). I'll fish out my old papers on all this, and we can get back to funning

I don't know whether to stick with original stats or not (?). Don't think it will make much difference, but do we stick with original data set or update it ? I don't mind - the principle's the same.

Great to have you back JavaHead, and much kudos to you.