Overview: Python visualization tools

[dhavide]

This requires notebooks to illustrate skills needed for the data visualization in the tutorials. This could be one or more notebooks (it could be about 20-30 minutes of teaching time for a live audience depending on prior experience).

• Focus on examples that use geospatial data
• Try to minimise the total number of different functions/methods invoked (i.e., assume that novice learners will need to have the particulars of any function explained to them if they've never seen it before).
• To minimise the number of different packages, try to use HvPlot/Holoviz primarily; it is possible that we'll have to use Matplotlib or Folium for some examples, but that will involve more overhead for learners.

[dhavide]

Still trying to decide which tools to focus on for plotting visualization (see discussion in #43 for more context).

• Matplotlib is robust but leads to verbose, clunky code
• Holoviews/HvPlot/Geoviews leads to cleaner code but is less robust
• Currently exploring Folium as an option.

[dhavide]

Things to remember to explain in whatever framework we settle on:

• Colormaps: how to customize in useful ways for our contexts (images with < 10 different pixel values in dynamic range)
• Transparency: how to ensure certain pixel values are transparent (see this discussion)
• Overlaying images / masks onto geographic background tiles
• Transforming geographic coordinates into image coordinates & visa versa
• Incorporating axis labels, etc.

Ideally the framework we choose should allow these concerns to dealt with automatically (i.e., by default). If not, we'll have to address that explicitly in the narrative for this notebook.