This talk demonstrates how to make data easily visualizable, build dashboards in notebooks and as deployable apps, plot millions or billions of data points in a web browser, and create a readable, maintainable, reproducible workflow.
Data visualization represents a core component of data science - from initial exploration, to detailed analysis, and then reporting and dissemination. The Python ecosystem has developed a plethora of packages for visualizing data, each making a trade off between customizability and ease of use.
Part of the tension between approaches has to do with the competing goals of visualization. Initial exploration needs to be quick and flexible, allowing users to explore and adjust any aspect of the data so that the most important features can be discovered. Once the key aspects have been identified, the data scientist might prepare a specific image or figure to share with colleagues or a wider audience. Or, they might need to set up an interactive way to share a set of data that would be unwieldy as a fixed figure, using interactive controls to let others explore the effects of certain variables.
In this talk we will build up visualizations starting from a quick first pass and ending at a fully deployable dashboard, using a set of open-source packages that have been designed to work well together. In particular, the audience will learn to:
All of the packages that will be discussed are BSD-licensed libraries that are part of the PyViz initiative (pyviz.org).