Scribd
Upload
8 views29 pages

Human-In-The-loop Data Science With Jupyter Widgets Presentation

Uploaded by

mutashima
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views29 pages

Human-In-The-loop Data Science With Jupyter Widgets Presentation

Uploaded by

mutashima
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 29

ASI Data Science is a

London-based data science


consultancy
About me

— Committer for Jupyter widgets


— Main author of jupyter-gmaps, a library for
visualizing geographical data in Jupyter notebooks
— Author of Scala for data science (Packt Publishing)
Jupyter widgets
for human-in-the-loop data
science
Developing machine
learning software is hard
Developing machine learning software is hard

— almost always stochastic


— often black box
Developing machine learning software is hard

— robustness
— overfitting
— overreliance on certain features or groups of features
Traditional software
development workflows are
inadequate
Human intuition

It should be easy for humans to explore the model.


Human intuition

Computers think in terms of bytes and instructions, and


humans think in terms of concepts and images.
Human intuition

We need a framework to rapidly create UIs that allow


the human to think at a higher level of abstraction.

The UI should not be a black box.


Jupyter widgets
Jupyter widgets allow building user interfaces entirely in
Python, directly in Jupyter notebooks.
Build user interfaces in
Python
directly in
Jupyter notebooks
Examples
Jupyter widgets

— Jupyter widgets are written entirely in Python


— They are written in the environment the data
scientist is currently working in
— Widgets have access to the entire state of the
notebook
Ecosystem
Core: ipywidgets
bqplot
ipyleaflet and gmaps
qgrid
ipyvolume
Learning about widgets

— mlviz: visualising machine learning algorithms with


Jupyter widgets and bqplot.
— Jupyter widgets tutorial
— Coding a simple widget from scratch: video and code
— Jupyter widgets documentation
Libraries used in this talk

— ipywidgets
— bqplot
— gmaps
— lens
— superintendent
Use widgets to
reduce friction
at the human computer
interface
Acknowledgements

— Jupyter widgets developers: Jason Grout, Sylvain


Corlay, Maarten Breddels, Matt Craig, Vidar Tonaas
Fauske
— ASI Data Science
— Chakri Cherukuri (ChakriCherukuri)
— Victor Zabalza (zblz) and Scott Stevenson (srstevenson)
— Jan Freyberg (janfreyberg)
— SherlockML

You might also like