What are the key differences?

It uses poetry rather than conda to package and manage dependencies.

I've created a command line to run this from. See the line in the toml file:

[tool.poetry.scripts]
zipfs-law = "zipfs_law.cli:main"

cli contains the workflow dag that creates the analysis.

I've separated most of the input and output into separate functions, as that big script was quite big.

Workflow

I've downloaded these samples in the data directory.

Now run at a terminal that has poetry activated:

zipfs-law data output

Open the notebook (jupyter notebook) and run all cells to check that the empirical distributions roughly look like the theoretical ones.