How do you make Jupyter notebooks reproducible?

medium

Answer

$Notebooks are anti-reproducible by default: out-of-order execution, hidden state.$
$Fixes: (1) Papermill: parameterize + execute notebooks as scripts programmatically.$
$(2) jupyter nbconvert --execute fresh run.$
$(3) nbdev / nbdime for diff + review.$
$(4) Extract library code into .py; notebook is thin runner.$
$(5) nbstripout git filter to remove outputs from commits.$
$(6) nb-clean on commit.$
$Rule: notebooks for exploration, .py for production.$

Check yourself — multiple choice

Random
Notebooks anti-reproducible by default; fix: Papermill parameterize + execute fresh + nbdev/nbdime diff + extract to .py library + nbstripout / nb-clean strip outputs; notebooks explore, .py produce
Just save
Not real

Notebook reproducibility: Papermill + fresh execute + .py library + strip outputs.

#reproducibility

215 interview questions in this topic.