Thanks for the strings migration — prerequisites, method table, to_titlecase / to_lowercase, extract, and the split → list.to_struct().unnest() pattern look good, and CI is green. A few fixes before merge:

Must-fix

1. Lost “last token after split” example
   Original taught split().str.get(-1) (last name). That was replaced with:

   df.with_columns(pl.col("names").str.slice(0, 5))

   which teaches something different. Please restore the last-token idea with polars list access, e.g.:

   df.with_columns(pl.col("names").str.split(" ").list.get(-1).alias("last"))

   Keep a slice() example if you want, but don’t drop the split→get teaching point.

2. get_dummies → column split needs clearer prose
   Original str.get_dummies(";") produced one-hot indicator columns. The new code:

   .str.split("; ").list.to_struct(...).unnest("tags")

   splits into fixed country / subject columns instead. That’s fine for this dataset, but the text should say so explicitly (and not imply it’s the same as dummy encoding). Optionally show a polars-native dummy approach later (explode + pivot / to_dummies) if you want to keep that skill.

Medium / nits

3. Grammar: “we can use the `str.split()` to split” → “use `str.split()` …”
4. Kernelspec display_name: ".venv" → generic Python 3 (ipykernel) (same note as other PRs).
5. Commit author metadata is still Your Name — please set git user.name/email for future commits.

Happy to approve after the last-token example (and clearer dummies→split prose) are sorted.