[Data] [3/N] StandardScaler Preprocessor with arrow format#59906
Merged
alexeykudinkin merged 28 commits intoray-project:masterfrom Jan 16, 2026
Merged
[Data] [3/N] StandardScaler Preprocessor with arrow format#59906alexeykudinkin merged 28 commits intoray-project:masterfrom
alexeykudinkin merged 28 commits intoray-project:masterfrom
Conversation
The GIL makes checking s`elf._serialize_cache is not None` atomic, so we don't need lock. Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com>
Signed-off-by: xgui <xgui@anyscale.com>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request introduces a performant, Arrow-native transformation path for the OneHotEncoder preprocessor, which is a great enhancement. The implementation correctly uses PyArrow and vectorized NumPy operations for efficiency. The new tests are comprehensive, especially the parameterized tests that ensure consistency between the pandas and Arrow implementations. I have one suggestion to improve a test case to make it more robust and accurate.
Signed-off-by: xgui <xgui@anyscale.com>
Signed-off-by: xgui <xgui@anyscale.com>
Signed-off-by: xgui <xgui@anyscale.com>
xinyuangui2
changed the title
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
iamjustinhsu
reviewed
Jan 8, 2026
Signed-off-by: xgui <xgui@anyscale.com>
Signed-off-by: xgui <xgui@anyscale.com>
goutamvenkat-anyscale
approved these changes
Jan 15, 2026
alexeykudinkin
approved these changes
Jan 16, 2026
Comment on lines
+137
to
+138
| # Read all input columns first to avoid reading modified data when | ||
| # output_columns[i] == columns[j] for i < j |
Contributor
There was a problem hiding this comment.
Comment is confusing -- there's no actual reading, you just get a reference to pa.ChunkedArray
limarkdcunha
pushed a commit
to limarkdcunha/ray
that referenced
this pull request
Jan 18, 2026
…ct#59906) Adopt the arrow format for StandardScaler preprocessor instead of pandas. Before (master with pandas): | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.556 | 180 | 5.56 | 5.49 | 5.97 | 6.47 | 5.35 | 6.73 | | StandardScaler | Arrow | 5 | 500 | 0.547 | 914 | 5.47 | 5.42 | 5.72 | 6.18 | 5.30 | 7.13 | | StandardScaler | Arrow | 10 | 1,000 | 0.549 | 1,821 | 5.49 | 5.42 | 5.60 | 7.36 | 5.33 | 7.73 | | StandardScaler | Arrow | 20 | 2,000 | 0.546 | 3,663 | 5.46 | 5.44 | 5.65 | 5.76 | 5.33 | 6.14 | | StandardScaler | Arrow | 50 | 5,000 | 0.555 | 9,016 | 5.55 | 5.41 | 5.78 | 6.69 | 5.32 | 13.40 | | StandardScaler | Arrow | 100 | 10,000 | 0.540 | 18,507 | 5.40 | 5.39 | 5.53 | 5.64 | 5.29 | 6.09 | After: | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.054 | 1,840 | 0.54 | 0.53 | 0.58 | 0.62 | 0.52 | 0.75 | | StandardScaler | Arrow | 5 | 500 | 0.055 | 9,041 | 0.55 | 0.54 | 0.63 | 0.78 | 0.53 | 0.79 | | StandardScaler | Arrow | 10 | 1,000 | 0.056 | 18,013 | 0.56 | 0.54 | 0.61 | 0.67 | 0.53 | 0.75 | | StandardScaler | Arrow | 20 | 2,000 | 0.055 | 36,653 | 0.55 | 0.53 | 0.58 | 0.62 | 0.53 | 0.77 | | StandardScaler | Arrow | 50 | 5,000 | 0.055 | 90,670 | 0.55 | 0.54 | 0.59 | 0.61 | 0.52 | 0.91 | | StandardScaler | Arrow | 100 | 10,000 | 0.055 | 181,200 | 0.55 | 0.54 | 0.59 | 0.62 | 0.52 | 0.78 | --------- Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com> Signed-off-by: xgui <xgui@anyscale.com> Signed-off-by: Limark Dcunha <limarkdcunha@gmail.com>
jeffery4011
pushed a commit
to jeffery4011/ray
that referenced
this pull request
Jan 20, 2026
…ct#59906) Adopt the arrow format for StandardScaler preprocessor instead of pandas. Before (master with pandas): | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.556 | 180 | 5.56 | 5.49 | 5.97 | 6.47 | 5.35 | 6.73 | | StandardScaler | Arrow | 5 | 500 | 0.547 | 914 | 5.47 | 5.42 | 5.72 | 6.18 | 5.30 | 7.13 | | StandardScaler | Arrow | 10 | 1,000 | 0.549 | 1,821 | 5.49 | 5.42 | 5.60 | 7.36 | 5.33 | 7.73 | | StandardScaler | Arrow | 20 | 2,000 | 0.546 | 3,663 | 5.46 | 5.44 | 5.65 | 5.76 | 5.33 | 6.14 | | StandardScaler | Arrow | 50 | 5,000 | 0.555 | 9,016 | 5.55 | 5.41 | 5.78 | 6.69 | 5.32 | 13.40 | | StandardScaler | Arrow | 100 | 10,000 | 0.540 | 18,507 | 5.40 | 5.39 | 5.53 | 5.64 | 5.29 | 6.09 | After: | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.054 | 1,840 | 0.54 | 0.53 | 0.58 | 0.62 | 0.52 | 0.75 | | StandardScaler | Arrow | 5 | 500 | 0.055 | 9,041 | 0.55 | 0.54 | 0.63 | 0.78 | 0.53 | 0.79 | | StandardScaler | Arrow | 10 | 1,000 | 0.056 | 18,013 | 0.56 | 0.54 | 0.61 | 0.67 | 0.53 | 0.75 | | StandardScaler | Arrow | 20 | 2,000 | 0.055 | 36,653 | 0.55 | 0.53 | 0.58 | 0.62 | 0.53 | 0.77 | | StandardScaler | Arrow | 50 | 5,000 | 0.055 | 90,670 | 0.55 | 0.54 | 0.59 | 0.61 | 0.52 | 0.91 | | StandardScaler | Arrow | 100 | 10,000 | 0.055 | 181,200 | 0.55 | 0.54 | 0.59 | 0.62 | 0.52 | 0.78 | --------- Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com> Signed-off-by: xgui <xgui@anyscale.com> Signed-off-by: jeffery4011 <jefferyshen1015@gmail.com>
ryanaoleary
pushed a commit
to ryanaoleary/ray
that referenced
this pull request
Feb 3, 2026
…ct#59906) Adopt the arrow format for StandardScaler preprocessor instead of pandas. Before (master with pandas): | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.556 | 180 | 5.56 | 5.49 | 5.97 | 6.47 | 5.35 | 6.73 | | StandardScaler | Arrow | 5 | 500 | 0.547 | 914 | 5.47 | 5.42 | 5.72 | 6.18 | 5.30 | 7.13 | | StandardScaler | Arrow | 10 | 1,000 | 0.549 | 1,821 | 5.49 | 5.42 | 5.60 | 7.36 | 5.33 | 7.73 | | StandardScaler | Arrow | 20 | 2,000 | 0.546 | 3,663 | 5.46 | 5.44 | 5.65 | 5.76 | 5.33 | 6.14 | | StandardScaler | Arrow | 50 | 5,000 | 0.555 | 9,016 | 5.55 | 5.41 | 5.78 | 6.69 | 5.32 | 13.40 | | StandardScaler | Arrow | 100 | 10,000 | 0.540 | 18,507 | 5.40 | 5.39 | 5.53 | 5.64 | 5.29 | 6.09 | After: | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.054 | 1,840 | 0.54 | 0.53 | 0.58 | 0.62 | 0.52 | 0.75 | | StandardScaler | Arrow | 5 | 500 | 0.055 | 9,041 | 0.55 | 0.54 | 0.63 | 0.78 | 0.53 | 0.79 | | StandardScaler | Arrow | 10 | 1,000 | 0.056 | 18,013 | 0.56 | 0.54 | 0.61 | 0.67 | 0.53 | 0.75 | | StandardScaler | Arrow | 20 | 2,000 | 0.055 | 36,653 | 0.55 | 0.53 | 0.58 | 0.62 | 0.53 | 0.77 | | StandardScaler | Arrow | 50 | 5,000 | 0.055 | 90,670 | 0.55 | 0.54 | 0.59 | 0.61 | 0.52 | 0.91 | | StandardScaler | Arrow | 100 | 10,000 | 0.055 | 181,200 | 0.55 | 0.54 | 0.59 | 0.62 | 0.52 | 0.78 | --------- Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com> Signed-off-by: xgui <xgui@anyscale.com>
peterxcli
pushed a commit
to peterxcli/ray
that referenced
this pull request
Feb 25, 2026
…ct#59906) Adopt the arrow format for StandardScaler preprocessor instead of pandas. Before (master with pandas): | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.556 | 180 | 5.56 | 5.49 | 5.97 | 6.47 | 5.35 | 6.73 | | StandardScaler | Arrow | 5 | 500 | 0.547 | 914 | 5.47 | 5.42 | 5.72 | 6.18 | 5.30 | 7.13 | | StandardScaler | Arrow | 10 | 1,000 | 0.549 | 1,821 | 5.49 | 5.42 | 5.60 | 7.36 | 5.33 | 7.73 | | StandardScaler | Arrow | 20 | 2,000 | 0.546 | 3,663 | 5.46 | 5.44 | 5.65 | 5.76 | 5.33 | 6.14 | | StandardScaler | Arrow | 50 | 5,000 | 0.555 | 9,016 | 5.55 | 5.41 | 5.78 | 6.69 | 5.32 | 13.40 | | StandardScaler | Arrow | 100 | 10,000 | 0.540 | 18,507 | 5.40 | 5.39 | 5.53 | 5.64 | 5.29 | 6.09 | After: | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.054 | 1,840 | 0.54 | 0.53 | 0.58 | 0.62 | 0.52 | 0.75 | | StandardScaler | Arrow | 5 | 500 | 0.055 | 9,041 | 0.55 | 0.54 | 0.63 | 0.78 | 0.53 | 0.79 | | StandardScaler | Arrow | 10 | 1,000 | 0.056 | 18,013 | 0.56 | 0.54 | 0.61 | 0.67 | 0.53 | 0.75 | | StandardScaler | Arrow | 20 | 2,000 | 0.055 | 36,653 | 0.55 | 0.53 | 0.58 | 0.62 | 0.53 | 0.77 | | StandardScaler | Arrow | 50 | 5,000 | 0.055 | 90,670 | 0.55 | 0.54 | 0.59 | 0.61 | 0.52 | 0.91 | | StandardScaler | Arrow | 100 | 10,000 | 0.055 | 181,200 | 0.55 | 0.54 | 0.59 | 0.62 | 0.52 | 0.78 | --------- Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com> Signed-off-by: xgui <xgui@anyscale.com> Signed-off-by: peterxcli <peterxcli@gmail.com>
peterxcli
pushed a commit
to peterxcli/ray
that referenced
this pull request
Feb 25, 2026
…ct#59906) Adopt the arrow format for StandardScaler preprocessor instead of pandas. Before (master with pandas): | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.556 | 180 | 5.56 | 5.49 | 5.97 | 6.47 | 5.35 | 6.73 | | StandardScaler | Arrow | 5 | 500 | 0.547 | 914 | 5.47 | 5.42 | 5.72 | 6.18 | 5.30 | 7.13 | | StandardScaler | Arrow | 10 | 1,000 | 0.549 | 1,821 | 5.49 | 5.42 | 5.60 | 7.36 | 5.33 | 7.73 | | StandardScaler | Arrow | 20 | 2,000 | 0.546 | 3,663 | 5.46 | 5.44 | 5.65 | 5.76 | 5.33 | 6.14 | | StandardScaler | Arrow | 50 | 5,000 | 0.555 | 9,016 | 5.55 | 5.41 | 5.78 | 6.69 | 5.32 | 13.40 | | StandardScaler | Arrow | 100 | 10,000 | 0.540 | 18,507 | 5.40 | 5.39 | 5.53 | 5.64 | 5.29 | 6.09 | After: | Preprocessor | Scenario | Batch Size | Total Rows | Total Time (s) | Throughput (rows/sec) | Avg Latency (ms) | P50 (ms) | P95 (ms) | P99 (ms) | Min (ms) | Max (ms) | |--------------|----------|------------|------------|----------------|----------------------|------------------|----------|----------|----------|----------|----------| | StandardScaler | Arrow | 1 | 100 | 0.054 | 1,840 | 0.54 | 0.53 | 0.58 | 0.62 | 0.52 | 0.75 | | StandardScaler | Arrow | 5 | 500 | 0.055 | 9,041 | 0.55 | 0.54 | 0.63 | 0.78 | 0.53 | 0.79 | | StandardScaler | Arrow | 10 | 1,000 | 0.056 | 18,013 | 0.56 | 0.54 | 0.61 | 0.67 | 0.53 | 0.75 | | StandardScaler | Arrow | 20 | 2,000 | 0.055 | 36,653 | 0.55 | 0.53 | 0.58 | 0.62 | 0.53 | 0.77 | | StandardScaler | Arrow | 50 | 5,000 | 0.055 | 90,670 | 0.55 | 0.54 | 0.59 | 0.61 | 0.52 | 0.91 | | StandardScaler | Arrow | 100 | 10,000 | 0.055 | 181,200 | 0.55 | 0.54 | 0.59 | 0.62 | 0.52 | 0.78 | --------- Signed-off-by: Xinyuan <43737116+xinyuangui2@users.noreply.github.com> Signed-off-by: xgui <xgui@anyscale.com> Signed-off-by: peterxcli <peterxcli@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adopt the arrow format for StandardScaler preprocessor instead of pandas.
Before (master with pandas):
After: