Machine learning in trading: theory, models, practice and algo-trading - page 2124

New comment
 
mytarmailS:

How Ivakhnenko recommends dividing, so that the model is trained properly

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content393/Content393.htm

this is not enough, because later you may get new data from a totally different distribution

I used to do such things long ago in old bots

 
Maxim Dmitrievsky:

this is not enough, because then you may have new data from a completely different distribution

I did that a long time ago in old bots

I agree ...

By the way, it's a cool book, it's like a Soviet school, and everything is there, wood, networks, rsa, modeling, hypotheticals and in clear Russian

 
mytarmailS:

How Ivakhnenko recommends dividing, so that the model is trained properly

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content393/Content393.htm

Recommended literature: [ Ivakhnenko, 1969, 1982; Ivakhnenko and Lapa, 1971; Ivakhnenko et al., 1976; Brusilovsky, 1987; Ivakhnenko and Jurachkovsky, 1987; Rosenberg et al., 1994 ].

? topical reading ))))

Ibid, beginning of the opus:

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content0/Content0.htm#Ref

An analysis of a variety of memes that are close in construction by V.P. Leonov [URL= http://www.biometrica.tomsk.ru/lis.htm - (Uniform Resurse Locator) - Uniform Resource Locator; this is how we will mark references in the list of references, presented by addresses in the Internet without the year of publication indicated], confirms the ideas expressed by V. V. Nalimov [1989] about probabilistic distribution of meanings. The following traditional transformations of memes in the scientific environment can be distinguished:

  • Being on the same medium, heterogeneous memes cause an increased probability of generating fundamentally new combinations, which can potentially form into a constructive scientific hypothesis (the principle of "cross-pollination by ideas" announced at the Davos Forum);

  • in the course of semantic engineering there appear memes with a complex construction that do not always allow a sufficiently accurate and reliable representation of the meaning of the concepts and notions that the authors try to describe with their help (in their language and style such descriptions often resemble the famous works of Andrei Platonov "Kotlovan" and "Chevengur");

  • a dreary (albeit harmless) picture is presented by long caravans of dogmatic memes, having the form of appeals to the Central Committee of the CPSU on the occasion of a socialist holiday and representing magical incantations, serving, according to the authors, as a "pass to science" (almost no ecological work does not go without mentioning the complex nature of the interaction of biocenosis components or the need for a systems approach to their study, although often the authors make not the slightest attempt to highlight this complexity or systematically study it)

  • sometimes there is a semantic rupture of a concept, after which separate parts of the term, like viruses, begin to live an independent life, transferring sections of information they carry to other contexts, and the meaning of the former part is assigned to the meaning of the new whole (this may include the process of generating new terms under which purely speculative concepts are hidden, and the rediscovery under a new appearance of regularities that have been used in other branches for a long time, and endless "terminological wars");

  • "camouflage memes" are meaningless and absurd expressions that are combinations of terms sounding and incomprehensible to the authors - clichés borrowed from other publications and needed to achieve the main goal - to give the work scientific "weight" (the use of computers and statistics often becomes a "ritual camouflage" component, designed to artificially raise the importance and weight of the work).




are you really reading this?

 
mytarmailS:

I agree...

By the way, it's a cool book, the Soviet school as it is, and everything is there, both wood and nets and rsa and in clear Russian

awesome book, for beginners in the MO it's the best

 
Igor Makanu:

? topical reading ))))

there , the beginning of the opus:

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content0/Content0.htm#Ref


are you really reading this?

http://gmdh.net/articles/theory/bookInductModel.pdf

a big plus is that linear models always converge to a local minimum. That's why the method is still relevant

 
Igor Makanu:

? topical reading ))))

there , the beginning of the opus:

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content0/Content0.htm#Ref

are you really reading this?

what's wrong?

Maxim Dmitrievsky:

it's a great book, for beginners in the iO it's perfect

well
 
mytarmailS:

How Ivakhnenko recommends dividing, so that the model is trained properly

http://www.ievbras.ru/ecostat/Kiril/Library/Book1/Content393/Content393.htm

This will not work for timeseries. It's an analogy to mixing a train with a test. There will be a peek at the dots next to each other.

 
elibrarius:

It doesn't work for timeseries. It is an analogy of mixing a trayn with a test. There will be a peek at the points next to each other.

If you remove the autocorrelation of the traits, it's fine.

 
elibrarius:

It doesn't work for timeseries. It is an analogy of mixing a trayn with a test. There will be a peek at the dots next to each other.

Yes I understand, but the point itself is good, separate not just by statistical properties and evenly across the test and trail

 
Maxim Dmitrievsky:

If you remove the autocorrelation of the traits, it will work.

If all points from both the test and the trace are ranked in one common list (rearranged according to some pattern), it means that they are mixed up. My understanding is this. The test should not be mixed in any way with the trail.

1...211721182119212021212122212321242125212621272128212921302131...3516
New comment
Reason: