Machine learning in trading: theory, models, practice and algo-trading - page 579

[Dmitriy Skub]

Yuriy Asaulenko:

Would you be so kind as to drop me a line? If this is the same torrent that I have, there is a version for mechanical engineering.

There are also versions for science, but it does not say about the tablet.

It turns out there is the entire suite).

I threw.

[Maxim Dmitrievsky]

Yuriy Asaulenko:

A question for anyone who might know.

There is only one tree. What kind of boundary will it build in N-dimensional space?

There's more than one boundary... if you look at 2-dimensional space, it divides it into ranges - rectangles. Well for n-dimensional it would be hyper phygogons

picture from Wikipedia

[Yuriy Asaulenko]

Maxim Dmitrievsky:

there is more than one boundary... if you look at 2-dimensional space, it divides into ranges - rectangles. Well for n-dimensional it would be hyper phygogons.

wikipedia picture.

Is this for one tree?

One tree - one line on the plane?

[Maxim Dmitrievsky]

Yuriy Asaulenko:

Is this for one tree?

I think one tree is one line on the plane? for a 2D picture. or am I wrong?

No, one split is one line... i.e. splitting the sample into two parts...

it may end up looking like these intricate images, not like the NS line

[Yuriy Asaulenko]

Maxim Dmitrievsky:

No, 1 split is 1 line... i.e. splitting the sample into 2 parts...

it might end up looking like these intricate images, not like the NS line

Is this picture for a single tree? I'm interested in RF from a single tree right now. I need to understand for a single tree. I'll figure it out on my own from here.

What I read, it's not clear. It seems to be clear, but only for the aggregate.

I understood that one tree gives a hyperplane in N-dimensional space. It turns out I got it all wrong.

[Maxim Dmitrievsky]

Yuriy Asaulenko:

Is this picture for a single tree? I'm interested in RF from one tree right now. I need to understand for one tree. I'll figure it out on my own.

What I read, there is nothing clear. It seems to be clear, but only for the aggregate.

yes, this is for decision tree - the decisive tree

the committee of trees is then assembled into a random forest model via bagging (bootstrap aggregating)

according to the principle - a lot of simple trees will give better results than a complex model

RF is a bunch of regular trees assembled through bagging.

[Maxim Dmitrievsky]

Yuriy Asaulenko:

I understood that one tree gives a hyperplane in N-dimensional space. Turns out I got it all wrong.

plane is a linear classification, trees can not do it (or worse than linear regression), only non-linear

[Yuriy Asaulenko]

Maxim Dmitrievsky:

yes, this is for decision tree - the decision tree

then a committee of trees is assembled into a random forest model via bagging (bootstrap aggregating)

according to the principle - a lot of simple trees will give better results on average than a complex model.

RF is a bunch of regular trees collected through bagging.

If you have a link to something intelligible - pdf any, throw in a private message please. Please don't use Hubr.)

I would like something big and detailed.)

[Maxim Dmitrievsky]

Yuriy Asaulenko:

If you have a link to anything intelligible - pdf any, throw in the personal pliz. Don't do it on Habr.)

I would like something big and detailed).

I don't remember, I just googled a few articles.

like https://basegroup.ru/community/articles/description

there everything is simple

[Yuriy Asaulenko]

Maxim Dmitrievsky:

I don't remember anymore, I read several articles on google

here for example https://basegroup.ru/community/articles/description

Thank you. I wonder if there are any monographs, exist in the nature?