18.4 Project and Process

Q16. Walk through a recent analysis.

I load the extract, normalise columns, fix types and duplicates, explore with groupby/crosstab, plot daily trend and city GMV, then write recommendations. In this course project I intentionally stop before machine learning.

Q17. What is grain and why does it matter?

Grain is what one row represents (e.g. order line vs order). Counting rows at line grain overstates order counts; I use nunique on order_id when needed.

Q18. How do you make analysis reproducible?

Notebook + venv + saved clean CSV + savefig charts + markdown insights. Same inputs should regenerate the same outputs.

Q19. Excel/Power BI vs Python — when Python?

When cleaning steps must be repeated weekly on new extracts, when I need custom metrics/plots in code, or when the workflow belongs in version control. Dashboards for interactive slicing may still belong in Power BI.

Q20. Do you know machine learning?

I can discuss concepts at a high level, but this course focused on analytics-strength Python: cleaning, EDA and visualisation. I treat ML as a separate skill path.