PyViz: Easy Visualization and Exploration for all your Data | SciPy 2018 Tutorial | James A. Bednar
Vložit
- čas přidán 11. 07. 2018
- In this tutorial, you will learn how to use the PyViz suite of tools to quickly build simple or complex visualizations that reveal and give insight into your data.
You'll start by using the HoloViews library to annotate your numpy, pandas, or xarray data and make it have an instantly available visual representation. These declarative objects make it easy to visualize how different sets of data relate to each other, by flexibly overlaying and laying out data in any combination. Once you have these objects, you can select, slice, or sample your data as needed, quickly making new types of plots so that you can fully understand your data. You'll be able to choose between Matplotlib plots suitable for publications, or Bokeh plots that support interactive exploration in Jupyter Notebooks or as separate dashboard apps.
You'll then see how to use HoloViews and related tools to transform your data for viewing in different ways, defining complex analysis pipelines if needed that preserve the raw data while making each step visible for analysis. When your datasets are too large for web browsers, we'll show how to use Datashader to reveal all of the data faithfully, rendering it into an image that can be displayed safely even for billions of datapoints. When it is time to share your results with non-developer colleagues, we'll show how to use Param to add interactive widgets to let them explore parameter spaces, how to link plots to respond to user selections and events, and how to put plots and widgets together into a separately launchable dashboard.
Throughout, we will demonstrate how to customize plots as needed, how to follow up on specific topics to get more information, and how and when to use each of the libraries described. The aim is to give users the tools and know-how to effectively explore, analyze and visualize even large and complex datasets easily, concisely, and reproducibly. You can see a detailed breakdown of the topics included, including the full training materials as Jupyter notebooks, at pyviz.github.io/pyviz/tutoria....
See tutorial materials here: scipy2018.scipy.org/ehome/299...
See tutorial materials here: scipy2018.scipy.org/ehome/299...
See the full SciPy 2018 playlist here: • SciPy 2018: Scientific... - Věda a technologie
great overview of my favorite library.
Where can I access the notebook? It looks like the tutorial materials link in the description is dead
Hi, all materials seem to be corrupted, I get unreadable notebook error in jupyter, can someone take a look, thanks! Great tutorials!
Does all of this work with Jupyterlab ?
Please use the 2019 version at czcams.com/video/7deGS4IPAQ0/video.html instead of this one; the tools are even better now!
Will there be a 2019 version of this talk?
Yes; see czcams.com/video/7deGS4IPAQ0/video.html
What jupyter extension allows the full screen/presentation of a notebook like that?
github.com/damianavila/RISE
Thanks! I saw you called that out later in the video - it was just 45min after seeing it haha. How long does it take you to format a notebook for RISE? I only had one formatted, so I assume it's non-trivial.
To make a notebook into RISE slides, just select for each cell whether it is a slide or a fragment of the previous slide. But then you have to make the formatting work at the larger size, by breaking up long lines of code, so I only do it for some notebooks.
conda install -c pyviz pyviz channel not available please help
It's now `conda install -c pyviz holoviz`.
yet there's 300 MM views on Cardi B's new song.