PyLadies Seattle - Lessons in Interactive Visualizations
1 like275 views
The document discusses interactive visualization techniques in Python, particularly in the context of environmental data such as the Elwha River. It emphasizes the importance of critical thinking regarding the ethics and politics of data visualization and outlines various Python libraries to create interactive plots. The speaker expresses gratitude towards the open source community and the Pyladies Seattle for their support and participation in the event.
PyLadies Seattle - Lessons in Interactive Visualizations
- 1. INTERACTIVE PLANETARY lessons in interactive python vis amcasari pyladies seattle, 23feb2016 nasa
- 2. COORDINATES who: @amcasari what: @PyLadiesSEA where: @Concur why: #Pythonistas #PyLadiesRock (now we can be found)
- 3. data science via random walks senior data scientist @ Concur control systems engineering + robotics + legos officer in USN operations research analyst wandering dirtbag + conservation volunteer EE + applied math + complex systems underwater robotics consultant extraordinaire SAHM
- 4. STORYTELLING LOOP DA LOOP You Experience Show Tell They Experience
- 5. TRADITIONAL ANALYSIS + STATIC VISUALIZATION: WE MOSTLY TELL 2 + 2 = 4
- 6. HOW DO WE SEE THE WORLD: EXPERIENCE the elwha river, olympic peninsula, wa
- 7. HOW DO WE SEE THE WORLD: SHOW the elwha river, olympic peninsula, wa
- 8. HOW DO WE SEE THE (DATA) WORLD: TELL then there’s this… seattle times: the elwha river, olympic peninsula, wa american rivers nps
- 9. HOW DO WE REALLY, REALLY SEE THE WORLD? field biologist
- 10. You Experience INTERACTIVE VISUALIZATION! Tell n.b. not approved by Concur’s design team :) Show User’s Experience
- 11. W/ INTERACTIVE VISUALIZATION: WE SHOW + THEY EXPERIENCE
- 12. “because i can…ethics. bias. data != numbers. call to think critically about the politics + ethics of visualization - catherine d’ignazio the point of collection - mimi onuoha nasa
- 13. HOW DO WE (THOUGHTFULLY) SHOW OUR (DATA) WORLD? periscopic: global gender gap in phds nytimes: what does a marriage license cost you 2016 primary results + calendar xkcdsmartmine whale tracker
- 14. HOW DO WE (THOUGHTFULLY) SHOW OUR (DATA) WORLD? all figs from papers here show + tell != show + experience
- 15. HOW DO WE SHOW OUR (DATA) WORLD WITH PYTHON?
- 16. HOW DO WE SHOW OUR (DATA) WORLD WITH PYTHON?
- 17. HOW DO WE SHOW OUR (DATA) WORLD WITH PYTHON?
- 18. HOW DO WE SHOW OUR (DATA) WORLD WITH PYTHON?
- 19. + + ➤ choose your install: conda, pip ➤ seaborn> python viz library based on matplotlib ➤ ipywidgets> add interactive HTML widgets to Jupyter notebooks ➤ which takes advantage of matplotlib’s interactive backend connectors…. ➤ and allows you to layer seaborn for pretty, interactive, lightweight vis ➤ extensible stack for any Python visualization library! ➤ Jupyter + ipywidgets + {insert your fav here} ➤ deploy using Jupyter nbviewer ➤ easy to piece things together for reproducible analysis: reusable code + interactive plots + deployable notebooks
- 20. # note to self: boh-kay, not boh-kah ➤ choose your adventure: python, julia, scala, r ➤ choose your install: conda, pip ➤ targets web browsers for presentation ➤ uses tornado to create a bokeh server output_file() ➤ can also plot in jupyter notebook w/o bokeh server output_notebook() ➤ uses mplexporter to help convert matplotlib(!) plots in Bokeh plots ➤ plays well with others: seaborn, ggplot.py, pandas ➤ excellent documentation ➤ recent updates are making way for “BIG DATA” vis ➤ brilliant deep dive into bokeh with christine doig of continuum analytics
- 21. + } ➤ choose your install: conda, pip ➤ uses HTML’s SVG…so not really designed for “BIG DATA” vis ➤ still developing! noted missing features include: ➤ tick locations + tick formatting ➤ plt.xkcd() ➤ plt.annotate() ➤ excellent + growing documentation, including faq’s ➤ extensible beyond matplotlib > client-side interface is pure javascript library (n.b. current JSON specification designed for matplotlib) n.b. this is a jake vanderplas project
- 23. UNDERSTANDING YOUR INTERNET WORLD export bookmarks (.html) BeautifulSoup my_links my_dirs my_names bag o’words still getting clever str, regex http_v_https counts all the vis DEMO!?!
- 24. HOW DO WE SHOW OUR (DATA) WORLD WITH PYTHON? ➤ Embed in your favorite Python web framework ➤ Django ➤ Flask ➤ Tornado ➤ Pyramid ➤ Jupyter Notebooks ➤ nbviewer: standalone server for hosting notebooks ➤ Jupyter + github :) ➤ n.b. github renders only static plots (non-html) on jupyter notebooks ➤ Pay us to host your vis! ➤ plot.ly
- 25. ➤ graphlab create is based on a python data science library developed + (some) os’d by dato ➤ graphlab canvas: interactive visualization for exploratory data analysis NOT ALL BITS ARE OS….BUT….
- 26. ➤ python library developed + os’d by stitchfix ➤ pyxleyJS React to create Flask-based web apps “Through the use of the PyReact library, we can use Jinja templating to construct and transform a single React component. The specific UI components are passed as props to the parent component. A simpler interface is provide through the use of specific wrappers for each of the component types.” AND EVEN SHINIER?
- 27. {THANKS MUCH} ➤ thank you to everyone in the open source community for giving me such lovely tools to talk about ➤ thank you @PyLadiesSEA for listening ➤ thank you again to @Concur for hosting, snacks + being a fantastic place to be a PyLady ➤ slides + git repo links will be posted to meetup page amanda.casari@concur.com @amcasari