Machine learning in trading: theory, models, practice and algo-trading - page 583

New comment
 
Yuriy Asaulenko:

What makes you think that? Where did you get it from?

I do not use signs for the system. With signs I only cut off from the time series (and from training and from functioning) the areas where it is not necessary to analyze anything at all.

The NS itself directly chews the time series.

I've already written and even cited a book -NS performs well for highly specialized tasks in combination with conventional methods.


Sorry, you wrote about 15 inputs to NS... I automatically associated it with the number of signs :)

 
Maxim Dmitrievsky:

Oh, I'm sorry, you wrote about 15 entrances to the NS... I automatically associated it with the number of signs :)

Well, here's a quote from the book. I've already cited it 3-4 times here.))


That's where I started from.

By the way, after 2 days of study, it seems to me that the woods are not very well suited to solve market problems (well, of course, it also depends on the formulation of the problem).

To tell the truth, I thought the same thing about NS for a long time).

 
Yuriy Asaulenko:

Well, here's a quote from the book. I've already quoted it here 3-4 times.)

That's where I got it from.


I can't get the picture to load.

is it from this edition?

 
Maxim Dmitrievsky:

I can't get the picture to load.

is that from this edition?

Yes, it is.

I have the picture in my post.

 
Yuriy Asaulenko:

Yes, from here.

I have the picture in my post.

I also can not see the picture at once, and now is not visible.

 
Konstantin Nikitin:

Also, the picture was not visible at once, and now is not visible.

Reloaded the picture.
 
Yuriy Asaulenko:

By the way, after 2 days of study, it seems to me that the forests are not very good for solving market problems (well, of course, it still depends on the formulation of the problem).


And why forests are not suitable?

p.s. Yes, in the picture everything is correct, an integrated approach... that's what I'm trying to do :) not "stupid" to teach the model

 
Maxim Dmitrievsky:

Why aren't scaffolds suitable? They solve the same problems

I'm trying to put a time line in there. And there are simply no attributes for forests in it. Forests with predictors is more or less clear, but in the time series there are hardly any sets shared by forests. (And overkill on the predictors is not our method.)
 
Yuriy Asaulenko:
I'm trying to put the time series there. And there are simply no features for forests. Forests with predictors is more or less understandable, but time series hardly has sets that are shared by forests. (And overkill of predictors is not our method.)

I agree, we need an automatic feature extraction at least :)

 
Maxim Dmitrievsky:

I agree, we need automatic feature extraction at least :)

So the NS is such an extractor. By structure itself, NS is a set of filters. That is, NS itself is a set of self-organizing predictors.
1...576577578579580581582583584585586587588589590...3495
New comment
Reason: