Time series forecasting with Deductor Academic 5.2 - page 6

New comment
 
Reshetov:

This DA is quite weak.

I ran it on a simple classic example of recognition:

a regular one (although the statistics are much better of course)... but grids are grids anyway.... you just need to tweak the settings (it won't work on complex data of course) ....

did it for fun ... single layer 6 neur.... found everything you asked for....

 
Vizard:

ordinary (although the statistics are much better of course)... but grids are grids anyway.... you just need to tweak the settings (it won't work on complex data of course) ....

did it for fun ... single layer 6 neur.... found everything you asked for....

Yeah figured out the settings already. The Deductor's help is probably the worst I've ever seen. You can gradually figure out from the interface what's what, but it takes time and patience. Otherwise, you can say that the horse is working, if you can manage it.
 
Reshetov:
Deductor's helpdesk is probably the worst I've ever seen....
that's for sure...
 
Of the Artificial Intelligence programs, the Ded is superior to absolutely all Western counterparts, such as: Although Ded is not a masterpiece DAT-MINIG, there are more brutal in processing analytics and the end result of the program, and they are the same Russian. For example this one-- ProfiStat Lite. Many "chips" were already present here 2-3 years ago, and grandpa has a shitty logical regression and decision tree appeared only 1 year ago. ProfiStat is easy to understand even for a child, but there are no hacked programs on the Internet.
 
I am often asked how to run a logical regression, I answer: after pre-processing the data should be loaded into "column convolution" block, where in the column "transposable" 2 values should be added, the first of which should be date-time, the second price value (Klos, Open, whatever) and in "informational" only price values should be thrown . ... The main drawback is not timely adaptation to changing market conditions, the linear regression ( generally it's a rubbish). Although the forecast can reach fantastic accuracy for 2 - 3 days in advance. But not being able to adapt crosses its value in the end.
 
AAAksakal:
The Artificial Intelligence program is superior to all Western analogues, such as SAS, STATISTICS, ORACUL, SPSS: SAS, STATISTICS, ORACUL, SRSS, these programs are suitable for accountants from Zimbabwe, and you can not call them artificial intelligence programs at all. The West lags Russia in this field, because education in the West in fundamental areas such as physics and mathematics is at an extremely low level ... Although Grandpa is not a masterpiece DAT-MINIG, there are more brutal for analytical processing and end results, but they are also Russian . For example this one-- ProfiStat Lite. Many "chips" were already present here 2-3 years ago, and grandpa has a shitty logical regression and decision tree appeared only 1 year ago. ProfiStat is handled even by a child, but there is no hackable program on the Net.


Grandpa is not even close to statistics.... stop joking ))))))))

It's made just for Zimbabwe - serious analysts and forecasters don't even pay attention to such micro-programs...

 
Statistics is a plebeian program, I only mentioned it because it is popular and completely useless. It is capable of counting seeds in a sack rather than analysing and drawing conclusions.
 
Vizard mate cross the US department, BOING, MITSUBISHI, prefer Russian artificial intelligence software.
 
Half a year ago, you asked me about PolAnlit, I still do not understand it and do not intend to, that is, the data processing download is all without problems, but I could not get a forecast, over there Reshetov is smart, maybe he will figure it out, but dumb apparently, so if you were on the website PolAnlit, you saw a list of companies and countries that use it. Granddad is not Anal, but it is made by us.
 
AAAksakal:
Half a year ago, you asked me about PolAnlit, I still do not understand it and do not intend to, that is, data download, processing, all without problems, but I could not get a forecast, von Reshetov our smart, maybe he understands, but dull apparently, so if you've been to the site PolAnal, you saw a list of companies and countries that use it. Grandfather not Anal cannot be compared, but it is made by us.

Six months ago I told you how to use polyanalyzer (run the data - get the formula - use it) + note that the version that is on the net is a bit glitchy and not complete....

.... and why don't you get the forecast you hoped for from the deductor or anywhere else.... (i.e. price forecast) ... the polynalyst is also a product of the Russian brain by the way ... Reshetov is smart (fools don't code their grids in 2 days) - but he couldn't make a forecast straight away... this once again confirms grandpa's weakness... although grids are by and large the same everywhere.... ! in the same statistics their choice (according to grid structures) is much larger +committees, selection by condition + many different gadgets....

BOING, MITSUBISHI...etc... don't watch who uses what... everyone's tasks are different etc....

stats is great... and much, much better than grandpa... (it and other big packages are used by serious analysts and forecasters and some programmers interested in forex.... i didn't get it from the ground up, i just crossed paths with people)

https://www.mql4.com/go?http://s58.radikal.ru/i161/1008/cc/3ed02ac2f935.jpg
https://www.mql4.com/go?http://s005.radikal.ru/i209/1008/d9/a68271cc13af.jpg

but it's a bit complicated and inconvenient (in my opinion + I'm as dumb as a stick and unlearned) .... but i use it sometimes...

for other products date mineral - the availability of not 1-2 but 10-20 models of arim for example (i do not remember the name of the program exactly) does not say that it is better, etc etc....

but for the most part just need to set realistic goals then everything will be found......

1234567
New comment