Page **3** of **4**

### Re: Statistical data gather for you

Posted: **May 4th, 2019, 12:23 pm**

by **MarkBrezina**

That's quite nice. I'm a Dane as you all know, I live in Copenhagen, have a polish girlfriend living in London and I study physics, on the side I work as a tutor in a variety of subjects for highschoolers.

I also agree, Finland is too far away(so is Iceland) from the rest of Scandinavia. I guess we're all clumped together since the Vikings

### Re: Statistical data gather for you

Posted: **May 4th, 2019, 8:01 pm**

by **katastrofa**

I think this here is the most interesting until now.

The first is basically politics/news/current affairs

Second is maths

Companies/markets/finance

Dealing with people

IT

I'm wondering why the big prevalence of mathematical modelling tasks over programming if every models ends up as code. Maybe people qualify both as math modelling to avoid ticking the box listing programming together with such intellectually inferior professions as IT?

Actually, there's an ongoing debate in my field whether the code is the model. Journals turn to the policy of requiring the code to be revealed - a problem for guys like I, who build proprietary models (with their numerical implementations). I have to agree with that to some extent, though. I would even venture to say that in some fields, especially ML, the computer architecture becomes an integral part of the model - it's necessary to reproduce the result. As such omni-models grow - allowing to analyse more complex problems in more complex ways - it becomes increasingly harder to embrace all their elements. A simple example: dynamic execution (processor executes instructions in an random order) may affect insignificant digits in conventional modelling, but may change the result in reinforcement learning.

Even my, former physicist's mind starts to wonder whether the old Popperist method (i.e. falsification, which ruled the past century's science) is not obstructing the way of further scientific progress (a problem already encountered in cosmology, which struggles to test the theories of universe(s)). In many fields we are now capable of modelling phenomena at such a level of complexity and detail, that maybe we can no longer assume that the results, in which we are interested, are reproducible and "fundamental" - they are by the analysed problem's nature random and short-lived. Right as I hoped we were already past postmodernism.

### Re: Statistical data gather for you

Posted: **May 4th, 2019, 8:32 pm**

by **MarkBrezina**

I agree, modelling in general could be argumented as done by either part in the development of an analysis strategy. A mathematician or a programmer/IT architect. I think that would be depending on which person is actually shaping the model and in part also in the people building the model. As somebody of course could produce a basis idea and everyone could develop on that basis idea.

My initial reason for doing all the statistics I'm doing now on the financial system, was that I couldn't decide on which method to use for stocks analysis, I had something like 30 methods which I really liked. But when I came here however I saw even more useful and interesting methods. So I gave up that game entirely until I had done the basis work.

I imagine the sad story in much of this is that a lot of it ends up like the logistic map(chaos iteration equation?) Initial conditions are of grave importance for a reproducible or understandable result.

On the end, I sense a question coming along so I will do my best to answer it.

I think you will find your answer is in the repetitions of time, but the timespan needed is immense and the answer will perhaps be unintelligeble in the way that you won't be capable of pronouncing a simple phrase to describe such a truth. If more detail is added and you simply further the complexity to keep all the details in place. The system you're building will slowly blow up. Even this is a repetitive truth from itself onto itself. Thereby a repitition of a ... Yeah I'll stop here.

### Re: Statistical data gather for you

Posted: **May 5th, 2019, 8:09 pm**

by **Cuchulainn**

A nice result would be the half-life of a thread, a bit like a Markov model, e.g. how many responses, their frequency, and duration etc. And the drift.

Actually, MIS

https://en.wikipedia.org/wiki/Managemen ... ion_system

### Re: Statistical data gather for you

Posted: **May 5th, 2019, 8:45 pm**

by **MarkBrezina**

Alright you are thinking about the statistical data I'm gathering for this forum right?

Using it to look at half-life of threads, number of responses, response frequency and thread lifetime?

MIS to do that? how about excel ark?

### Re: Statistical data gather for you

Posted: **May 5th, 2019, 10:30 pm**

by **MarkBrezina**

Alright fellas time for the next update on my little one man project.

I've now had 40 responses on my survey

### Re: Statistical data gather for you

Posted: **May 5th, 2019, 10:31 pm**

by **MarkBrezina**

Last one is on what information sources people use and I'm imagining this could be of quite a lot more significance, so I'll wait with that one

### Re: Statistical data gather for you

Posted: **May 6th, 2019, 10:36 am**

by **Cuchulainn**

*Actually, there's an ongoing debate in my field whether the code is the model. *

No, it is not 1:1.

The model is technology-independent, while code is technology-dependent, 1:N It is a product/value chain in a process of elaboration.

Your field thinks in 1:1 terms because the concept of software product does not exist in general. But you can't hold it against them as they have not been trained to think as software designers.

This can be a yuge culture shock when newly-minted PhDs enter a product-design team.

### Re: Statistical data gather for you

Posted: **May 6th, 2019, 3:37 pm**

by **katastrofa**

The phrasing is meant just to express the general concept.

@"The model is technology-independent, while code is technology-dependent" - hence my po-mo rambling (if the model is code-dependent, it may also turn out to be architecture dependent, and eventually so fragile to factors which influence its deployment that it can't reproduce its own results).

### Re: Statistical data gather for you

Posted: **May 6th, 2019, 11:17 pm**

by **MarkBrezina**

I sincerely do believe the model might be made mostly by the one who shapes the idea, and that might actually be a single person in a firm

### Re: Statistical data gather for you

Posted: **May 7th, 2019, 12:49 pm**

by **Cuchulainn**

I sincerely do believe the model might be made mostly by the one who shapes the idea, and that might actually be a single person in a firm

Wishful thinking.

### Re: Statistical data gather for you

Posted: **May 15th, 2019, 7:19 pm**

by **MarkBrezina**

Heya fellas, sorry that I'm not around these days, my courses has startet again and my students have their exams next week, so I'm running up a couple of hills these days. I'll probably be back when everything has settled a bit

### Re: Statistical data gather for you

Posted: **May 15th, 2019, 9:59 pm**

by **katastrofa**

Good luck with everything! Watch the saddle points.

### Re: Statistical data gather for you

Posted: **May 15th, 2019, 11:52 pm**

by **ISayMoo**

*Actually, there's an ongoing debate in my field whether the code is the model. *

No, it is not 1:1.

The model is technology-independent, while code is technology-dependent, 1:N It is a product/value chain in a process of elaboration.

Models are rarely technology-independent if your dataset size is unlimited.

### Re: Statistical data gather for you

Posted: **May 16th, 2019, 2:15 pm**

by **Cuchulainn**

Good luck with everything! Watch the saddle points.

No one talks about bifurcation?