Visualise a decision tree in Colaboratory

Question

What is the best way to visualise a decision tree using Google Colab? The visualisations from 'dtreeviz' (e.g.Github) are really neat, but when running something like

!pip install dtreeviz

and

from sklearn.datasets import *
from sklearn import tree
from dtreeviz.trees import *

followed by

classifier = tree.DecisionTreeClassifier(max_depth=4)
cancer = load_breast_cancer()
classifier.fit(cancer.data, cancer.target)
viz = dtreeviz(classifier,
              cancer.data,
              cancer.target,
              target_name='cancer',
              feature_names=cancer.feature_names, 
              class_names=["malignant", "benign"],
              fancy=False )  

viz.view()

I get

ExecutableNotFound: failed to execute ['dot', '-Tsvg', '-o', '/tmp/DTreeViz_62.svg', '/tmp/DTreeViz_62'], make sure the Graphviz executables are on your systems' PATH

Which could have something to do which Colab running via my g-drive?

Any help appreciated!

@MaximilianPeters, thank you - I tried it now. Still no visualisation returned - but also now no error message. — RandomForestRanger, Jun 18 '19 at 10:41
Hah, we are getting there! Can you try `from IPython.core.display import display, HTML; display(HTML(viz.svg())`? — Maximilian Peters, Jun 18 '19 at 19:07
Gosh, top man for persisting. And...You've solved it! I've added a ")" at the end of your code. — RandomForestRanger, Jun 20 '19 at 07:50

score 3 · Accepted Answer · answered Jun 20 '19 at 08:30

Short answer

Make sure graphviz is installed via !apt-get install graphviz
You can get the created SVG via viz.svg()
Wrap the output in IPython's HTML and then call display to show it in your notebook
```
from IPython.core.display import display, HTML
display(HTML(viz.svg()))
```

Longer answer

dtreeviz view() creates a SVG file in your temp directory
This file gets passed to the graphviz library which opens it depending on your OS
Google colab is recognized as linux and it tries to open the SVG file via the default viewing application
The last step leads to nowhere if you are not running the notebook locally (probably the Google server now has either a couple of open SVG images or some error messages)
The code from the short answer just gets the SVG code without saving it and then displays it in the notebook

Visualise a decision tree in Colaboratory

1 Answers1