DimensionalData.jl: named dimensions for julia data | Schouten | JuliaCon 2024

DimensionalData.jl: named dimensions for julia data by Rafael Schouten
PreTalx: pretalx.com/ju...
DimensionalData.jl is not the first or only package to define arrays and datasets with named dimensions and lookups in julia - AxisArrays.jl, AxisKeys.jl, NamedDims.jl all succeed in this space.
But DD is unique in its focus on extensibility, and has spawned many objects inheriting its behaviors, largely across spatial sciences, such as in Rasters.jl, YAXArrays.jl and ClimateBase.jl. It's also generic enough to find uses in statistics, machine learning and image processing.
Another focus driven by its spatial use-cases is on being able to accurately represent a wide rang of lookup values and behaviors:
Points or Intervals
Regular or Irregular
Ordered or Unordered
Sampled or Categorical
Linear or Cyclic
grid-aligned or with transformed coordinates
This means it can represent the wide variety of CF compliant netcdf files, as well as affine transformed coordinates and projections of GDAL.
DD also shines in ease of plotting, with extensive Plots.jl and Makie.jl recipes that automatically put the right values and labels in the right places.
This talk will demonstrate the basics of both how to use DimensionalData.jl, and for package developers - how to extend it.

[@nowanilfideme2]

Love the update since I as a Julia-curious user of xarray was checking it out.
Having any array wrapped is nice, because of sparsity and such.