Random forest with bootstrap = False in scikit-learn python

What does RandomForestClassifier () do if we select bootstrap = False?

As defined in this link

http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier

bootstrap: boolean, optional (default = True) Whether bootstrapping patterns are used when building trees.

Having asked this because I want to use an arbitrary forest approach to the time series, so train with a sliding window of size (tn) and a predicted date (t + k) and would like to know if this will happen if we select True or False:

1) If Bootstrap = True, then when training samples can be of any day and any number of functions. So, for example, it can have samples from day (t-15), day (t-19) and day (t-35), each with randomly selected functions, and then predict the date output (t + 1).

2) If Bootstrap = False, he will use all the samples and all functions from the date (tn) to t to learn, so he will actually follow the order of the dates (that is, he will use t-35, t-34, t-33 ... and etc. Up to t-1). And then it will predict the date output (t + 1).

, Bootstrap , Boostrap = False, ( ) t-39 t-19 t-15, t + 1. .

... Bootstrap?

+6
3

, . .

, . Bootstrap = False - , False, .

, max_features, max_features , ( , , ).

, , - RF. , , () .

+5

. . scikit-learn , , , bootstrap = True ( ). , bootstrap = FALSE, , - , .

+5

[1]

- , . , , bootstrap = True ( ).

:

, bootstrap = True ( ).

Bootstrap = True ( ): Bootstrap = False:

[2] . , , , .

, () .

bag of balls

№ 1, .

, , 2, " " (bootstrap = True)

But if you put Subset 1 aside and take M more balls from the bag for your subset 2, then none of the balls in subset 1 can be in subset 2 (or in any other subset), because you "draw without replacement" bootstrap = False)

[1] https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html

[2] http://methods.sagepub.com/Reference//encyclopedia-of-survey-research-methods/n516.xml

0
source

Source: https://habr.com/ru/post/1658206/


All Articles