TST: Benchmarks for CSV writing for various index variations #39318
Conversation
Benchmarks derived from GH37484. This captures a peculiar difference in timings between a DataFrame originating from `df.set_index(...).head(...)` and one from `df.head(...).set_index(...)`.
|
can you show these your PR (e.g. |
|
Yes, here's the benchmark output: |
asv_bench/benchmarks/io/csv.py
Outdated
| f"col{i}": np.random.uniform(0, 100000.0, ROWS) for i in range(COLS) | ||
| } | ||
| df = DataFrame({**index_cols, **data_cols}) | ||
| self.df_standard_index = df.head(N_HEAD) |
There was a problem hiding this comment.
the HEAD is a bit confusing here. can you not just make ROWS=N_HEAD to achieve the same?
There was a problem hiding this comment.
It's like that because the following two tests do take a head with a smaller number of rows, and I didn't want to duplicate the creation code. However, that's easily solved by extracting that into a function.
There was a problem hiding this comment.
oh it was fine, but ok on the change. ping when green.
|
Hello @DanielFEvans! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found: There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻 Comment last updated at 2021-01-21 16:25:08 UTC |
|
thanks @DanielFEvans |
Benchmarks derived from GH37484 (#37484). This captures a peculiar difference in timings between a DataFrame originating from
df.set_index(...).head(...)and one fromdf.head(...).set_index(...).I've not included a whatsnew entry, as no section seems particularly relevant to internal-only test/benchmark changes. If it should go in, shout and I'll add a note somewhere.