Машинное обучение в трейдинге: теория, модели, практика и алготорговля - страница 1077

[FxTrader562]

Maxim Dmitrievsky :
Now I think I can drop out 1-st 2 lines, because 3-d line do it all

But how are you going to drop out the lines?

Are you going to write the code for every feature ... I mean all the 100 features if we use 100 features :)))

[FxTrader562]

Maxim Dmitrievsky :

yes, I leave 1-st line and combine together 2 and 3, 3-d (now 2-nd) line will consist all 2 features variants

so 1-st step: we are choosing best predictors without polynomial transformations

    2-nd step: combining all with which others with different polynomials, by adding 3-d predictor, where its multiplication of 2 predictors with different degrees

and next - need to combine this predictors from step 2 to which others

Are you sure this line is correct? I mean the syntax is correct?

m[i].Set(2,MathPow(RDFpolicyMatrix[i][(int)models2[bf][1]],firstDegree)*MathPow(RDFpolicyMatrix[i][(int)models2[bf2][2]],secondDegree));
         

My brain already hurts by seeing these 2D matrix and all that stuff and so just checking if the syntax is correct:))))))

[FxTrader562]

Maxim Dmitrievsky:
nope, I will use all 3 lines, all good now )

But what next...:)) are you going to write 4th line, 5th line,....so on for all features??

[FxTrader562]

Maxim Dmitrievsky:

now we have: best triple features, transformed with polynomials

can combine all best features now, adding 6 inputs, next 12 inputs and so on

and then can choose the best model over all

No..I want to know if you are going to write the code again and again for every feature or will the existing loops and code will do it all?

[FxTrader562]

Maxim Dmitrievsky:

this strings now contain 1 unit, which takes 3 inputs

we can operate with this units now, multiply it, so it can takes 3+3+1 = 7 inputs

this is holy sh.t

Well, you decided to write it that way and till now I am not 100% sure as why you are trying to write in that way..:)))

I am not an expert in writing programs of 2D arrays or matrix and hence, here my role is limited....

I can only understand important concepts quickly and help you somewhat to simplify things little better...but you should know how to program it...:))))

[FxTrader562]

Maxim Dmitrievsky :

I just try with 3 lines now, fixing an errors

if the error is reduced by line

OKKK ,, ..... Now I am getting closer in understanding why you are trying to implement in this way .... :))

If you want, you can provide me with the full code and I can try different variants of polynomials and it by replacing in this line:

m [i] .Set ( 2 , MathPow (RDFpolicyMatrix [i] [( int ) models2 [bf] [ 1 ]], firstDegree) * MathPow (RDFpolicyMatrix [i] [( int ) models2 [bf2] [ 2 ] secondDegree) );

[FxTrader562]

Maxim Dmitrievsky :

now just some errors in heaps of loops .. ^))) I try to fix

You mean syntax errors???..I thought LOG LOSS errors:)))))))))))

[Sergey Chalyshev]

Господа, а не пошли бы Вы … в англискую часть форума?

[[Удален]]

Sergey Chalyshev:
Господа, а не пошли бы Вы … в англискую часть форума?

Поддерживаю.

Действительно, если хочется пообщаться на птичьем языке, откройте себе ветку, и гогочите себе там до упаду.

[[Удален]]

Здесь в этой ветке МО нет генератора идей. А нужны новые идеи. Вместо этого происходит чисто механическое обсасывание кем-то написанных пакетов.

Ветка сдохла.