Machine learning in trading: theory, models, practice and algo-trading - page 2479

New comment
 
iwelimorn #:

Thanks. Perhaps it's not agony, but my lack of fundamental knowledge.

Is this also true if multiple sets of independent variables correspond to a single variable?

Nit. This kind of thing is normal.
 
Dmytryi Nazarchuk #:

Did you even understand what you wrote?

Well, yes. This is called the inconsistency of the data. I've been dealing with networks for 20 years and can say I'm a keeper of this branch. And why do you ask?
 
iwelimorn #:

I agree with you, if the same example describes several states, we will get probability close to 1/n where n is the number of states when classifying by any available algorithm.

But there are no absolutely similar examples, they are similar to a certain degree. The question is how to detect this "similarity".


100 examples in three months on M5... I wonder... Do you select samples from the initial sample according to the rules that you then use in trading?

If two vectors are very close to each other but have different target values, then this forces the algorithm to make a small bend which leads to decrease of the model stability, when a minor change of the input vector leads to a considerable change of the result. This is also not good as the model becomes extremely sensitive to input data and therefore may make errors more often.

100 samples in 3 months on М5 is achieved by thinning the data, which is the basic strategy that makes you analyze the market not at every bar, but only at a certain point in time, when the condition for the analysis was formed. Read my article to approximately understand what I'm talking about. It's true that it is somewhat outdated and I do not use a lot of it (I've moved on), but the basic concept there has not changed!

 
Mihail Marchukajtes #:
Well, yes. This is called contradictory data. I've been dealing with networks for 20 years and I can say I'm a keeper of this branch. Why do you ask?

Have you been drinking again?)

Or what's more interesting? ))
 
Mihail Marchukajtes #:
Well, yes. This is called contradictory data. I've been dealing with networks for 20 years and I can say I'm a keeper of this branch. Why do you ask?
No. I just don't.
 
mytarmailS #:

Drinking again? :))

Or did you switch to something more interesting? ))
It's my day off, so I thought I'd have a chat. Teach the young 'uns some sense while I'm good.) (Again, I have no pipe, otherwise I would have got quite chatty :-(
 
Dmytryi Nazarchuk #:
No. Just no
Could you be clearer, because it's not quite clear what you mean. Or rather not at all clear :-)
 
Is there anyone here who knows how to parse?
 
Mihail Marchukajtes #:
Could you be clearer, because it is not quite clear what you mean. Or rather not at all clear :-)

When applying machine learning methods to RUNNING ROWS, the situation when the same set of input variables corresponds to the same dependent variable is practically never met. Different values of the dependent variable form a prediction error that must be minimized.

This entire thread is about minimizing prediction error, Axakal.

Plain Truths....

 
Dmytryi Nazarchuk #:

When applying machine learning methods to RUNNING ROWS , the situation in which

why random?

1...247224732474247524762477247824792480248124822483248424852486...3699
New comment