Visualizing a decision tree ( example from scikit-learn )

Question

I'm a noob in using sciki-learn so please bear with me.

I was going through the example: http://scikit-learn.org/stable/modules/tree.html#tree

>>> from sklearn.datasets import load_iris
>>> from sklearn import tree
>>> iris = load_iris()
>>> clf = tree.DecisionTreeClassifier()
>>> clf = clf.fit(iris.data, iris.target)
>>> from StringIO import StringIO
>>> out = StringIO()
>>> out = tree.export_graphviz(clf, out_file=out)

Apparently the graphiz file is ready for use.

But how do I draw the tree using the graphiz file? (the example did not go into details as to how the tree is drawn).

Example code and tips are more than welcomed!

Thanks!

---

Update

I'm using ubuntu 12.04, Python 2.7.3

Answer 1

Which OS do you run? Do you have graphviz installed?

In your example, StringIO() object, holds graphviz data, here is one way to check the data:

...
>>> print out.getvalue()

digraph Tree {
0 [label="X[2] <= 2.4500\nerror = 0.666667\nsamples = 150\nvalue = [ 50.  50.  50.]", shape="box"] ;
1 [label="error = 0.0000\nsamples = 50\nvalue = [ 50.   0.   0.]", shape="box"] ;
0 -> 1 ;
2 [label="X[3] <= 1.7500\nerror = 0.5\nsamples = 100\nvalue = [  0.  50.  50.]", shape="box"] ;
0 -> 2 ;
3 [label="X[2] <= 4.9500\nerror = 0.168038\nsamples = 54\nvalue = [  0.  49.   5.]", shape="box"] ;
2 -> 3 ;
4 [label="X[3] <= 1.6500\nerror = 0.0407986\nsamples = 48\nvalue = [  0.  47.   1.]", shape="box"] ;
3 -> 4 ;
5 [label="error = 0.0000\nsamples = 47\nvalue = [  0.  47.   0.]", shape="box"] ;
4 -> 5 ;
6 [label="error = 0.0000\nsamples = 1\nvalue = [ 0.  0.  1.]", shape="box"] ;
4 -> 6 ;
7 [label="X[3] <= 1.5500\nerror = 0.444444\nsamples = 6\nvalue = [ 0.  2.  4.]", shape="box"] ;
3 -> 7 ;
8 [label="error = 0.0000\nsamples = 3\nvalue = [ 0.  0.  3.]", shape="box"] ;
7 -> 8 ;
9 [label="X[0] <= 6.9500\nerror = 0.444444\nsamples = 3\nvalue = [ 0.  2.  1.]", shape="box"] ;
7 -> 9 ;
10 [label="error = 0.0000\nsamples = 2\nvalue = [ 0.  2.  0.]", shape="box"] ;
9 -> 10 ;
11 [label="error = 0.0000\nsamples = 1\nvalue = [ 0.  0.  1.]", shape="box"] ;
9 -> 11 ;
12 [label="X[2] <= 4.8500\nerror = 0.0425331\nsamples = 46\nvalue = [  0.   1.  45.]", shape="box"] ;
2 -> 12 ;
13 [label="X[0] <= 5.9500\nerror = 0.444444\nsamples = 3\nvalue = [ 0.  1.  2.]", shape="box"] ;
12 -> 13 ;
14 [label="error = 0.0000\nsamples = 1\nvalue = [ 0.  1.  0.]", shape="box"] ;
13 -> 14 ;
15 [label="error = 0.0000\nsamples = 2\nvalue = [ 0.  0.  2.]", shape="box"] ;
13 -> 15 ;
16 [label="error = 0.0000\nsamples = 43\nvalue = [  0.   0.  43.]", shape="box"] ;
12 -> 16 ;
}

you can write it as .dot file and produce image output, as showed in source you linked:

$ dot -Tpng tree.dot -o tree.png (PNG format output)

Answer 2

You were very close! Just do:

graph_from_dot_data(out.getvalue()).write_pdf("somefile.pdf")