Machine learning in trading: theory, models, practice and algo-trading - page 456

 
Vizard_:

At least six times I've spoken since the 4th forum, including a thread and clarification of the details of the latest version from the creator. No more.
Sorry, Teacher))) The post to the Doc tells you how to bury your game under the skirting board, but you still won't dig it...

I'm curious about the bugs, too. Write a blog post just to give a link later, than to say that I wrote it 6 times and won't do it again - search yourself if you need among hundreds of thousands posts on the forum. If you say A, then say B.

 
Aleksey Terentev:
Dear, before making any market forecasts you should have become acquainted with the essence of the problem. The foresight of time series is not a classification of plants and cancers by chips.


Michael, just one question.

Why are your data sorted in the timeseries dataset?


What is there to predict?
At least how to divide the dataset into a training and a test one?


This dataset is not for prediction, but for classification, in which the order of lines is not important...

 
Vizard_:

The posts are deleted, so they were addressed only to Misha and Yura. There are no mistakes on the books there...

So are there errors or not?

People interested in Reshetov's neuron still show up. I, for example, also fiddled with it and described the result in the blog. What prevents you from doing the same? Or it was not mistakes of neuron, but personal squabbles (if posts were deleted, it means they were not relevant)?

 
Vizard_:

The posts are deleted, so addressed only to Misha and Yura. There are no mistakes on the books there...


Where is this post??? poke the link....

 
Mihail Marchukajtes:

This dataset is not for prediction, but for classification, in which the order of lines is not important...

Okay. I'll explain it better.
Classification is a task of guessing object's class from a set of its attributes.
Regression reconstruction - the task of reconstructing Y from some X, which was absent in the training set.
Classification of time series - the task of predicting a class of an object by some set of its features that change over time.
Time series regression reconstruction - the task of reconstructing Y from some sequence X, which is time-bound and absent in the training set.


I think you're missing an important detail.

upd/ My point is that the dataset in machine learning is divided into training and test samples. If you divide that dataset, then there is nothing to train, because the whole test set will belong to one class, and the training set to another.
 
Vizard_:

I don't give a shit who shows up or who writes what.

If you can't explain your words about Reshetov's neuron having 2 errors, then your words are empty.
 
Elibrarius:
If you can't explain what you said about Reshetov's neuron having 2 errors, then your words are empty.

There is not and can not be any errors - it's just 1 quite ordinary neuron. Unless in the code, well, so, like, fix it).

I don't understand what he's talking about, either.

On the other hand, you can't do much with it (neuron). But Reshetov was one of the first to take up this topic.

 
Vizard_:

Another one))) It was written that I will not tell more, the limit is exhausted)))
Reshetov started writing this dud after one of my posts on 4. And I told Fomenko about dm, but he is old senile,
and I even nailed him after he'd been poking around with books, so he may not confirm it. Yura and Misha, as the only
who started using this rattlesnake, honestly told me at 4, about the shorts in the practical application, although the books are all true.
So you can say thank you that this rattle in the world at all))))


Thank you wizard and bow low (that was sarcasm if you did not understand at once) and wizard I intentionally wrote with a small letter.

Speaking of the latest version, I have it for example the 14th, and what release do you mean.... So Jura did not stand still, and had time to do a lot in the latest versions.....

I'm slowly getting into the Brown-Robinson method and the addition of this Brown-Robinson-Reshet method. Let's see what's up....

 

And let's define right away, I'm not talking about Reshetov's RNN, but about his optimizer... Not to be confused, these are completely different products.... that is... if anything....

 

Again...... I don't remember a polemic between Wizard and Reshetov, as the former claims. I must have missed it..... But the wizard is the king.... will be repeating to the lackeys what to say.... He has already said about it (referring to the mistakes) One more time to repeat not the king's blood....