As datasets become larger and more complex, visualizations tools that can handle this new scale of data are key in maintaining manageable workflows. For this tutorial, we will walk through a generalized dataviz-focused workflow that takes advantage of the performance of RAPIDS and RAPIDS-integrated libraries like cuDF, cuGraph, cuSpatial, cuxfilter, Datashader, hvplot with bokeh, and Plotly Dash.
The tutorial will be formatted around four key notebooks using a freely available dataset, based on a generalized visualization analysis workflow:
01 Data inspection and validation, using cuDF + hvplot
02 Exploratory data visualization, using cuDF + cuxfilter and datashader
03 Data analysis with visual analytics, using cuDF, cuSpatial and cuGraph + hvplot and datashader
04 Explanatory data visualization, using cuDF and cuGraph + Plotly Dash
Each notebook will be composed of incremental steps to prepare the data for visualization, choose the appropriate type of visualization, and apply basic analytics to reveal some interesting findings. Sections requiring user participation will have clear links to online resources and documentation. With Each notebook, accompanying video walkthroughs and screenshots have explanations of the process, so participants can compare their results with the video.