pandas str.rsplit + index is ~15.70× slower than polars str.split + list.get for splitting a string column

Splits on a hyphen and keeps the first piece.

Result

Run	pandas str.rsplit + index	polars str.split + list.get	Ratio
Run 1	459.83	29.29	15.70×
Run 2	458.4	30.29	15.14×
Run 3	564.18	34.35	16.42×

Spread over 3 runs: pandas str.rsplit + index CV 12.3% / polars str.split + list.get CV 8.6% / ratio CV 4.1%

df['s'].str.rsplit('-', n=1).str[0]

df['s'].str.split('-').list.get(0)

400,000 strings with two hyphens each.

Splitting into a list column and exploding it is a different operation.

cd bench && python run.py --run 1 --out ../data/run1.json