Houston we got a problem

On a sunny day (Thu, 18 Jun 2015 09:06:55 +0200) it happened Michael Welle wrote in :

OK, thanks. So, temperature maybe it can be ruled out, light, in the morning, could cause a ramp? Can your GM tube see any light, or is it enclosed?

the tube, the PSU and the electronics that hook the device to the network is built without any enclosure. To be precise it's located near a south facing window in the bright sun light.

I made a plot for the 06/05/2015, which was a hot and sunny day with temps > 34C. The graph looks more or less the same. The average CPM was

14.4 instead of 13.8 (for the 31/05/2015), but that's just noise I think.

Regards hmw

--
biff4emacsen - A biff-like tool for (X)Emacs 
http://www.c0t0d0s0.de/biff4emacsen/biff4emacsen.html 
Flood - Your friendly network packet generator 
http://www.c0t0d0s0.de/flood/flood.html

Or it can be selective interpretation. I don't see any pattern in this data. I see noise. In particular, notice that you can find the *same* sloped line in *many* places in the data. If such a pattern exists in the daily cycle, but is hidden by the noise, there are ways to show this. Just adding the data and looking at a noisy chart tells you little.

--

Rick

The chart indicates some periodic (one-day period) character, so the obvious place to start is a Fourier transform: find peaks in the frequency-domain, ignore the time-independent 'noise floor'.

So here's an answer with no four-dollar words.

I suggest that the data from the experiment does not meet expectations because of a problem with the expectations.

Clicky-event data arriving at low rates just naturally shows these kinds of variations.

Expectations are mis-calibrated because these kinds of data are not normally observed during day-to-day experience.

I think the unusual thing is why it doesn't vary more.

I liked your tritium experiment. Very cool.

:)

I'm reading a biography on Issac Newton by James Gleick. Newton had to invent his own four-dollar words. He started when he was a teenager. Now-days he might be forced into some kind of therapy.

--

Best Regards, 

ChesterW

So here's an answer with no four-dollar words.

I suggest that the data from the experiment does not meet expectations because of a problem with the expectations.

Clicky-event data arriving at low rates just naturally shows these kinds of variations.

Expectations are mis-calibrated because these kinds of data are not normally observed during day-to-day experience.

I think the unusual thing is why it doesn't vary more.

I liked your tritium experiment. Very cool.

:)

I'm reading a biography on Issac Newton by James Gleick. Newton had to invent his own four-dollar words. He started when he was a teenager. Now-days he might be forced into some kind of therapy.

--

Best Regards, 

ChesterW

Wow! Four dollar words. I guess that is inflation for you.

I think you are right. He found a data set that appeared to show a daily event. But this was done by eyeballing a graph that may not even have been prepared correctly. It would be better to do other math analyses to see just how much the data indicates what he thinks the data indicates.

--

Rick

Any random data set will have peaks. How do you proposed to distinguish the "noise floor" from any signal present?

--

Rick

If you're looking for a daily variation, cross-correlate the data with a 24h (or perhaps 23h:56m) signal. Perhaps autocorrelation would be more useful to find patterns, though.

On a sunny day (Thu, 18 Jun 2015 15:22:56 -0500) it happened ChesterW wrote in :

Statistics is not my field, and I am not a mathemagician. There was a professor in Germany who showed his students the way statistics can fool people. He found a city in Germany with the highest child birth was also the one with the most storks. So from that -statistically- it could be shown that storks bring children. This society and industry uses statistics - no - abuses- statistics no end like they used witchcraft in the medieval times.

Thanks, it is still running, who knows what will show up.

LOL I am at high age, and _personally_ I do not give a ... about what current science 'thinks'. There are things that are known and not made public, just like when they did the research foe the bomb in the US and found the periodic system was wrong. It took years for that to appear. A lot is classified, we will see it all being used in the next war. There is also a lot of what's the more than 4 letter word? Baloney? somebody re-invented the glow bulb with graphene?

For displays huh? Look at the efficiency.

10 ps? Maybe.. And of course a 100 % better battery is invented every few weeks.

Newton was cool, he also sometimes took the credit for what others did. It is possible the problem with him was caused when that apple fell on his head.

On a sunny day (Thu, 18 Jun 2015 15:17:58 -0500) it happened ChesterW wrote in :

To shed some light on the problem I bought 2 bananas yesterday. I ate one.

:-)

I see, if you don't understand it, it has to be nonsense.

Of course you know this because... uh, how exactly?

Now you are just bouncing off the walls.

Are you coming to rest yet?

--

Rick

Actually you are better off folding the data on itself with the correct period which is probably related to the real position of the sun (ie any variation correlated with the Earths orbital motion). Approximately every 24 hours but with small variations with time of the year.

There might be a recognisable difference between summer and winter due to the variation in the Earths orbital speed at aphelion and perihelion.

Fourier domain will show that you have peaks at certain frequencies. Maximum entropy done correctly will give you a more accurate result or complete nonsense if done badly. The OP might find this link a helpful practical introduction to periodograms:

Not perfect but it covers several of the common practical pitfalls.

--
Regards, 
Martin Brown

Witch burning came in with the Renaissance, not the Middle Ages.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

The noise expected in a counter of random-time events is the square root of the signal, roughly the square root of the number of events recorded. The RMS value of the noise is the same in the Fourier transformed data as in the original data. So, the noise expectation is calculable.

That response is orthogonal to the question.

--

Rick

No,it isn't. You can calculate the RMS noise values, and plot that with the data. Better, you can replace each data point with a vertical stroke, from (datavalue - RMS_noise_estimate) to (datavalue + RMS_noise_estimate)

On a sunny day (Fri, 19 Jun 2015 10:54:00 -0400) it happened Phil Hobbs wrote in :

Now that is something to look forward too in 'merrica :-)

More often its when you do understand it that it becomes clear its nonsense. Nonsense makes much of the world go round.

NT