Jupyter Notebook

Jupyter notebooks

What is Jupyter Notebook?

Jupyter
A web-based interactive development environment

Jupyter on Windows

If you use Anaconda it already comes with Jupyter notebook. On Windows you can run it from the start menu.

Alternatively:

Set up a virtual environment and install the following package:

pip install jupyter

Jupyter on Linux and OSX

Install

For Linux and OSX I recommend using virtualenv and installing with pip.

virtualenv -p python3 ~/venv3
source ~/venv3/bin/activate
pip install jupyter

Run

cd src/examples/jupyter/
jupyter notebook

Your browser should open. If not, there is a link in the terminal.

Jupyter New notebook

Create a new notebook (New / Python 3)
It is called "Untitled" - Rename it
Type in some code x = 2
Execute code either by pressing the arrow that looks like a triangle or by pressing Ctrl-ENTER.
Show content of variables. It is enough to enter the name of the variable in an edit box and "run that code".
import modules
Quit - shut down the notebook server.
Number is the execution order Out[] refers to the number of the cell for this this the output.

Jupyter Notebook file format

Jupyter notebooks are JSON file with a .ipynb extension. Open one such file with a plain text editor.
We can download file as Python (.py) exporting the code as a stand-alone python program.

Jupyter notebook edit and command mode

Modes: Blue - command mode, Green - edit mode
Enter - Switch to edit mode (or newline if already in edit mode)
ESC - Switch to command mode
Ctrl-Enter - execute current cell
Shift-Enter - execute current cell
Clear Cell data before saving
Button h for help
Button A - add cell Above current cell
Button B - add cell Below current cell
dd - delete current cell
...

Jupyter notebook Intellisense (TAB completition)

Just press TAB for word completition and method suggestions
When we typed in the name of an object and press dot . we get the list of available methods and attributes.

e.g.

import sys
sys.

name = "Foo"
name.

Jupyter add

Open an existing notebook: add.ipynb

Before commiting a Jupyter notebook in git, it is usually a good idea to remove all the generated content. So:

Edit / Clear Output of all Cells

Planets

The Planets example we saw in the Pandas chapter
planets.csv
planets.ipynb

Jupyter input

input.ipynb

File chooser

Avoid hard-coded pathes, use the FileChooser.
filechooser.ipynb

IPy Widgets

ipywidgets
Interact
Widget list
ipywidgets.ipynb

Jupyter Notebook and Markdown

Latex in Jupyter Notebook

To insert complex formulas.
latex.ipynb
Latex symbols

Jupyter StackOverflow

Download a dataset from the Stack Overflow survey.
unzip the file. Feel free to remove the __MACOSX/ directory if it is included.
SO/so.ipynb
SO/selected-rows.ipynb
SO/selected-columns.ipynb

Use SciPy image

scipy_image.ipynb

Create Random image

create_random_image.ipynb

Load image using OpenCV

opencv
opencv-python
load_image_using_opencv

Genes using Jupyter

jupyter notebook genes.ipynb

More Jupyter examples

generate.ipynb Generate a single Pandas DataFrame
salary_generate.ipynb
salary.ipynb
temperatures.ipynb
weather.ipynb
numpy_matrix.ipynb
seaborn_tips.ipynb

Jupyter Notebook: run on other port

jupyter notebook --port 8080

Jupyter Notebook: public IP

jupyter notebook --ip 192.168.1.10

Other Jupyter Notebook Kernels

Jupyter Kernels

Jupyter Notebook convert to other format (nbconvert)

nbconvert

jupyter nbconvert try.ipynb --to HTML
jupyter nbconvert try.ipynb --to PDF    (needs also pandoc)

Jupyter Notebook extensions

pip install jupyter_contrib_nbextensions
pip install jupyter_nbextensions_configurator
jupyter nbextensions_configurator enable --user

Scratchpad
Variable Inspector

Jupyter notebook Checkpoints

A very simple, one-level version control.
I would not rely on it as it is now. Use a real VCS. (e.g. git)

Jupyter notebook autoreload

def hello():
    print("hello")

examples/ml/reload.ipynb

Keyboard shortcuts

Jupyter Notebook