Notebooks, slides, and a fully interactive perspective-python and perspective-workspace example for my upcoming JupyterCon2020 Jupyter community: tools and practices talk.
Many data visualization libraries are built with static data in mind, where everything is known before the visualization is created. Analyzing data streams in real-time, however, is a crucial part of many industries. Combining JupyterLab’s ease-of-use and flexibility with Perspective, a high-performance streaming data visualization library, users can rapidly prototype, analyze, and visualize results from a multitude of data sources both live and static. Using real-time stock quotes from the IEX Cloud API, Junyuan Tan demonstrates how Perspective can be used to visualize streaming data, combine real-time information with analysis and data points from various sources, and export data snapshots using Apache Arrow, without leaving the Jupyter Notebook.
Attendees will learn:
- How to accumulate data from a streaming datasource in a Perspective table
- Visualizing, charting and analyzing data in real-time using PerspectiveWidget
- Chaining multiple Perspective tables to create your own augmented, streaming data sources
- Streaming, exporting, and storing data in the Apache Arrow format, and reintegrating that data back into Perspective
A working knowledge of Python is recommended but not required.
You'll need to have perspective-python installed, along with pyarrow==0.16.0, pandas, numpy, and tornado.
perspective_basics.ipynbcontains a walkthrough of the Perspective library.datasources.ipynbcontain my implementation of non-blocking, multi-process & multi-threaded datasources that interface with the IEX Cloud APIjupytercon.ipynbcontains the main talk notebook where I join the streaming and static datasources with Perspective.
To run the Tornado example, run python3 tornado_server.py and navigate to localhost:8888.