Your first notebook, and when a notebook is the wrong answer

1 · Learn the move · Template pattern

A notebook is the right first home for exploratory analysis: code, output, and your reasoning interleaved, one file. It is also a machine for manufacturing irreproducibility if you run cells out of order - the notebook remembers state your eyes cannot see, and cell 4 quietly uses a variable cell 9 redefined. The template pattern tames it: a fixed skeleton every notebook starts from - a header cell stating purpose, input files, and date; imports in one cell; sections in execution order; and a standing rule at the top: before anything from this notebook is believed, restart the kernel and run all cells clean. Know the exit, too: the moment a notebook is run routinely rather than explored - the same cells, every batch - it wants to be a script. Exploration is notebook country; production is not.

Generate my analysis notebook from the lab skeleton: cell 1 = header (purpose: [purpose], inputs: [files], author, date, and the rule "restart kernel and run all before trusting any output"); cell 2 = all imports; then sections [load / QC / analysis / figures], each opening with a markdown cell saying what it does and what it assumes from prior sections. No cell may depend on one below it. Add a final cell that prints the input filenames and row counts as a run receipt. If any part of my described workflow is repeated every batch unchanged, flag it as SCRIPT CANDIDATE rather than putting it in the notebook.