Machine learning in trading: theory, models, practice and algo-trading - page 893

 
Maxim Dmitrievsky:

I haven't spoken to you at all) thoughts go faster than I have time to write

It could all be done as an article, but I'm too lazy to do it now.

So do it... I'm not against it. Why do you have to bother the branch all the time...

 
Who can explain how to interpret a random forest? So that you can see the pattern with the nodes as in one decision tree.
 
forexman77:
Who can explain how to interpret a random forest? So that you can see the scheme with nodes, like in one decision tree.

Will this work? I googled it for 30 seconds with a laptop on a tired belly from hard work

https://aysent.github.io/2015/11/08/random-forest-leaf-visualization.html

https://shiring.github.io/machine_learning/2017/03/16/rf_plot_ggraph

https://towardsdatascience.com/random-forest-in-python-24d0893d51c0

Visualization of Leaf Nodes in Random Forests
Visualization of Leaf Nodes in Random Forests
  • aysent.github.io
Random Forests are among the most popular and best performing methods for general machine learning tasks as classification or regression. Random Forests combine several decision trees into a single ensemble (forest) and average predictions from individual trees. Each tree in the ensemble is trained on a random subsample on an input dataset...
 

Yes, I did not know that you are sometimes "Captain Obvious")

Finding out the optimal number of trees and signs has been seen many times, the whole internet is overflowing with it. And even more you can get an average of the importance of all the traits.

I, on the other hand, need to put all the trees into one and get all the nodes with thresholds.

 
forexman77:

Yes, I did not know that you are sometimes "Captain Obvious")

To find out the optimal number of trees and traits has been seen many times, the whole Internet is overflowing with it. And even more you can get an average of the importance of all the signs.

I, on the other hand, need to put all the trees into one and get all the nodes with thresholds.

And if you try to read the links?

 
Maxim Dmitrievsky:

How about trying to read the links?

I read it, maybe I didn't understand something. And by the way a lot of similar read and even managed to pull the results of each tree with nodes (leaves).

 
forexman77:

I read it, maybe I didn't understand something. And by the way, a lot of similar readings and even managed to pull the results with nodes (leaves) on all the trees in the forest.

What do you need in the end? There are diagrams with split trees... did I strain my guts for nothing?

 
Maxim Dmitrievsky:

What do you want in the end? There are diagrams with split trees... did I strain my gut for nothing?

What is shown in the picture is a diagram of a single tree.

As a result, you need to understand how the final result for the ensemble is built. I want to make averaged result of all trees and visualize it.

 
forexman77:

What is shown in the picture is a scheme of one tree.

As a result, you need to understand how the final result of the ensemble is built. That is to mold, as it were, from all the trees one average and of course visualize it.

I don't know how a forest can be visualized, the trees use different signs and subsamples

It seems to imply that you need to look at each individual tree.

I'll look again.

 
Maxim Dmitrievsky:

I don't know how a forest can be mapped, there are different signs and subsamples used in the trees

It seems to imply that you need to look at each individual tree.

I'll keep looking.

I searched for probably a week, reread. I haven't found anything that makes sense. There are some ideas, and then in the process of rethinking...

Maybe this was not intended to reveal everything completely?