3 Jupyter notebook (Python) tricks

Introduction

Recently, I've been taking good care of Jupyter notebook (Python3), I will post it because some tricks have increased.

environment

• Jupyter Notebook 4.2.x

• Python 3.5.x

• For example, it can be used in the Jupyter environment on Docker at https://github.com/mokemokechicken/jupyter-tensorflow.

Tips

Convert the result of the decision tree of sklearn into a figure and display it inline

Decision trees such as sklearn.tree.tree.DecisionTreeClassifier can output in .dot format, but trying to display them on a notebook is quite annoying. However, with certain preparations and procedures, you will be able to do that easily.

Things necessary

• With graphviz installed on the OS and the environment variable PATH in dot

Method

Define a function like this.

decision_tree_util.py

import os
from tempfile import mkstemp
import subprocess

from sklearn.tree.export import export_graphviz


def convert_decision_tree_to_ipython_image(clf, feature_names=None, class_names=None,
                                           image_filename=None, tmp_dir=None):
    dot_filename = mkstemp(suffix='.dot', dir=tmp_dir)[1]
    with open(dot_filename, "w") as out_file:
        export_graphviz(clf, out_file=out_file,
                        feature_names=feature_names,
                        class_names=class_names,
                        filled=True, rounded=True,
                        special_characters=True)

    from IPython.display import Image

    image_filename = image_filename or ('%s.png' % dot_filename)

    subprocess.call(('dot -Tpng -o %s %s' % (image_filename, dot_filename)).split(' '))
    image = Image(filename=image_filename)
    os.remove(dot_filename)
    return image

You can use it like this.

notebook

from sklearn import datasets
from sklearn.tree.tree import DecisionTreeClassifier
%matplotlib inline

iris = datasets.load_iris()
X = iris.data
Y = iris.target

clf = DecisionTreeClassifier(max_depth=3)
clf.fit(X, Y)
convert_decision_tree_to_ipython_image(clf, image_filename='tree.png')

Reload the imported file at runtime

You may want to reload because you changed the file you are importing from another location. If you write the following, it will be reloaded every time you run notebook. Before I knew this, I restarted the kernel every time (^^;

%reload_ext autoreload
%autoreload 2

Output multiple graphs, pandas.DataFrame, etc. on the notebook from the program

I want to output DataFrame etc.

from IPython.core.display import display

Import and

display(data_frame)

Then it will be displayed as usual. If you pass a character string etc., the character string will be output normally.

I want to output a graph

If you write like this,

notebook

from IPython.core.display import display
from numpy.random.mtrand import normal
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

display('graph1')
sns.distplot(normal(0, 1, size=1000))
sns.distplot(normal(0.1, 2, size=1000))

display('graph2')
sns.distplot(normal(-2, 2, size=1000))
sns.distplot(normal(2, 4, size=1000))

The graphs are frozen together and the graph comes to the bottom.

In that case, you should call plt.show () at the right time.

notebook

from IPython.core.display import display
from numpy.random.mtrand import normal
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

display('graph1')
sns.distplot(normal(0, 1, size=1000))
sns.distplot(normal(0.1, 2, size=1000))
plt.show()

display('graph1')
sns.distplot(normal(-2, 2, size=1000))
sns.distplot(normal(2, 4, size=1000))
plt.show()

This can be used not only for script on notebook, but also when you want to output to notebook from ordinary source code.

at the end

It was a trick.