Visualize a decision tree two different ways

Visualize a decision tree two different ways

Data School

2 года назад

8,403 Просмотров

Ссылки и html тэги не поддерживаются


Комментарии:

Data School
Data School - 29.07.2021 14:51

🌲 Decision trees are one of my favorite models because of their high interpretability! However, you can achieve much better performance by ensembling them in a Random Forest (at the expense of interpretability). Which do you prefer: Decision trees or Random Forests? 🤔

Ответить
Nitzan Meirom
Nitzan Meirom - 29.11.2022 19:47

I found this video super clear and helpful, thank you so much!

Ответить
andrene salmon
andrene salmon - 07.07.2022 03:50

Hi, thanks. Can you do a video on interpreting the export text in relation to the tree? Appreciated

Ответить
Bored Eggyolk
Bored Eggyolk - 22.11.2021 07:09

whoa thank you so much! i've been learning this for these past week. thank you for the knowledge!

Ответить
Kartikeya Sharma
Kartikeya Sharma - 28.09.2021 05:41

This is simply awesome! I hope you have a million subs soon.

Ответить
scign
scign - 09.08.2021 16:56

While you can't use this on a RandomForestClassifier model directly, you can use this method with each item of the "estimators_" list of the RF model (code below). Caution: running this on an RF model instantiated with the defaults will output 100 trees with unlimited depth. May take a minute to render!

rf = RandomForestClassifier()
rf.fit(X_train, y_train)
for dt in rf.estimators_: plot_tree(dt, feature_names = features, class_names=classes)

Ответить
atul singh
atul singh - 31.07.2021 07:41

Thanks Kevin

Ответить
Pearl321
Pearl321 - 30.07.2021 16:55

Thank you for uploading this video.
I've been struggling to visualize the result of Decision tree. this video helped me to do it

Ответить
J D
J D - 30.07.2021 06:32

Can you please show the relation between gini purity & entropy?

Ответить
Mod Hua
Mod Hua - 29.07.2021 22:15

Could you explain what data you are using to generate this tree. Also, if sex takes on 0 or 1, why you have <=0.5? Could you explain? Thanks

Ответить
Jaques D'Erasmo
Jaques D'Erasmo - 29.07.2021 19:57

Thanks a lot Kevin.

Ответить
A Beaumont
A Beaumont - 29.07.2021 16:47

Great.

Ответить
Phil Toa
Phil Toa - 29.07.2021 15:26

Nice.

Ответить
Satam Roy
Satam Roy - 29.07.2021 15:14

If only I could have opportunity for giving more than 1 👍

Ответить