Machine learning in trading: theory, models, practice and algo-trading - page 1864

[Nanix]

Maxim Dmitrievsky:

Thank you, brackets are still missing, for example, here

the red ones are as they should be.

mine does it all. Otherwise, the conditions will be crooked and the tree will count the wrong

Yes, it happens in a hurry. Here with if/else

else{
    if(L_30_1  <= 0.00044){ return 2;}
    else{
        if(L_25_1  <= 0.00047){
            if(L_5_1  <= 0.00012){ return 0;}
            else{ return 2;}}
        else{ return 2;}}}

[Mihail Marchukajtes]

I'm rooting for you guys. Show me the result when you figure out the brackets. I like the tree itself and it's interesting and simple in the way it's organized, which is what makes it so interesting. I'll keep my hand on the pulse....

[Renat Akhtyamov]

Mihail Marchukajtes:
I'm rooting for you guys. Show me the results after you figure out the brackets. I like the tree itself and it is very interesting and simple in the way it is organized and that is what makes it so interesting. I'm keeping my hand on the pulse of....

I'm interested, too.

but how it looks like in practice, I want to see a picture

just codes...

[Roman]

Renat Akhtyamov:

I'm interested, too.

But what does it look like in practice, I want to see a picture

I just want to see a picture...

Example of tree structure.
Like for real trees, there are nodes from which branches branch off.

[Renat Akhtyamov]

Roman:

An example of tree structure.
Like real trees, there are nodes from which branches branch off.

Thank you, it's starting to make sense now.

So we find the optimal variant?

Very similar to the PCB tracing algorithm.

[Mihail Marchukajtes]

Roman:

An example of tree structure.
Like real trees, there are nodes from which branches branch off.

I've shown you exactly the same way, it was just clearer to me in the code :-)

[Roman]

Renat Akhtyamov:

Thanks, now it's starting to make sense.

So we find the best option?

Very similar to the PCB trace algorithm.

Yes, the activation function selects this or that node, and so on through the tree structure, until the best option is found.

Here's another interesting animation that reminds me of this algorithm.

[Roman]

Mihail Marchukajtes:
I showed you exactly the same, it was just clearer to me in the code :-)

Some time ago, when I was grasping the meaning of OOP
While smoking on the balcony, in my mind, I laid out the trees standing in front of my window into classes. :))
Good training in understanding.
A tree is just a branching structure.

[Valeriy Yastremskiy]

Renat Akhtyamov:

Thanks, now it's starting to make sense.

So we find the best option?

Very similar to the PCB tracing algorithm.

Yes, everything started with graphs.)

[Renat Akhtyamov]

Roman:

Yes, the activation function selects this or that node, and so on through the tree structure, until the best option is found.

Here's another interesting animation that reminds me of this algorithm.

OK.

And here is Maxim's code,

https://www.mql5.com/ru/forum/86386/page1862#comment_17290073

looks like a digital filter to me personally