Skip to content

Update Xgboost Benchmarks #220

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
avolkov-intel wants to merge 4 commits into
base: main
Choose a base branch
from
Open

Update Xgboost Benchmarks #220

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2bb557b
Initial commit
avolkov-intel Aug 21, 2026
11849bd
Add bioresponse and nyc_taxi_green datasets, update xgboost configs, …
avolkov-intel Aug 24, 2026
319ebb2
Add task_dispatch option to avoid data duplication for airline_depdel…
avolkov-intel Sep 1, 2026
fc72160
Minor optimization
avolkov-intel Sep 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions configs/common/xgboost.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,8 @@
"device": "cpu",
"estimator_params": {
"tree_method": "hist",
"n_jobs": "[SPECIAL_VALUE]physical_cpus"
"n_jobs": "[SPECIAL_VALUE]physical_cpus",
"enable_categorical": true

@david-cortes-intel david-cortes-intel Sep 1, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is enabled by default in the latest versions. Which xgboost version are you using where it makes an effect?

@avolkov-intel avolkov-intel Sep 1, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We need to specify it to ensure comparison with previous xgboost version is correct

},
"enable_modelbuilders": false
}
Expand All @@ -16,7 +17,8 @@
"device": "gpu",
"estimator_params": {
"tree_method": "hist",
"n_jobs": "[SPECIAL_VALUE]physical_cpus"
"n_jobs": "[SPECIAL_VALUE]physical_cpus",
"enable_categorical": true
}
},
"data": { "format": "cudf" }
Expand Down
97 changes: 81 additions & 16 deletions configs/regular/xgboost_binary.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,11 +6,11 @@
"data": {
"dataset": "airline_depdelay",
"preprocessing_kwargs": {
"category_encoding": ["onehot", "ordinal"],
"subsample": 600000
"category_encoding": ["ordinal"],
"subsample": 700000
},
"split_kwargs": {
"train_size": 100000,
"train_size": 200000,
"test_size": 500000
}
},
Expand All @@ -26,37 +26,102 @@
},
{
"data": {
"dataset": "hepmass",
"dataset": "airline_satisfaction",
"preprocessing_kwargs": {
"category_encoding": ["ignore", "ordinal"]
},
"split_kwargs": {
"train_size": 200000,
"test_size": 1000000
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.1,
"reg_alpha": 1.0,
"max_depth": 6,
"reg_lambda": 1.0,
"max_leaves": 256,
"n_estimators": 500
"n_estimators": 1000,
"enable_categorical": true
}
}
},
{
"data": {
"dataset": "amazon_employee_access",
"preprocessing_kwargs": {
"category_encoding": ["ignore", "ordinal"]
},
"split_kwargs": {
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.05,
"max_depth": 6,
"reg_lambda": 1.0,
"n_estimators": 500,
"enable_categorical": true
}
}
},
{
"data": {
"dataset": "gisette",
"dataset": "apsfailure",
"preprocessing_kwargs": {
"replace_nan": ["ignore", "mean"]
},
"split_kwargs": {
"train_size": 2000,
"test_size": 5000
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.15,
"learning_rate": 0.05,
"max_depth": 6,
"subsample": 0.8,
"colsample_bytree": 0.8,
"reg_lambda": 1.0,
"n_estimators": 200
}
}
},
{
"data": {
"dataset": "bioresponse",
"split_kwargs": {
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.05,
"max_depth": 6,
"subsample": 0.8,
"colsample_bytree": 0.8,
"reg_lambda": 1.0,
"n_estimators": 200
}
}
},
{
"data": {
"dataset": "hepmass",
"split_kwargs": {
"train_size": 400000,
"test_size": 1000000
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.1,
"reg_alpha": 1.0,
"reg_lambda": 1.0,
"max_leaves": 256,
"colsample_bytree": 0.1,
"colsample_bynode": 0.1,
"n_estimators": 100
"n_estimators": 200
}
}
}
Expand Down
44 changes: 5 additions & 39 deletions configs/regular/xgboost_multi.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,35 +4,11 @@
"gbt multi classification data": [
{
"data": {
"dataset": "letters",
"split_kwargs": {
"train_size": 0.5,
"test_size": 0.5
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.05,
"reg_lambda": 1,
"max_depth": 6,
"n_estimators": 200
}
}
},
{
"data": {
"dataset": "mnist",
"split_kwargs": {
"train_size": 20000,
"test_size": 50000
}
"dataset": "connect"
},
"algorithm": {
"estimator_params": {
"max_leaves": 256,
"colsample_bytree": 0.2,
"colsample_bynode": 0.5,
"n_estimators": 100
"n_estimators": 500
}
}
},
Expand All @@ -42,20 +18,10 @@
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.1,
"learning_rate": 0.3,
"reg_lambda": 1,
"max_depth": 8,
"n_estimators": 200
}
}
},
{
"data": {
"dataset": "connect"
},
"algorithm": {
"estimator_params": {
"n_estimators": 500
"max_depth": 14,
"n_estimators": 100
}
}
}
Expand Down
86 changes: 58 additions & 28 deletions configs/regular/xgboost_regression.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,17 +4,25 @@
"gbt regression data": [
{
"data": {
"dataset": "twodplanes",
"dataset": "airline_depdelay",
"dataset_kwargs": {
"task": "regression"
},
"preprocessing_kwargs": {
"category_encoding": "ordinal",
"subsample": 700000
},
"split_kwargs": {
"train_size": 0.33,
"test_size": null
"train_size": 200000,
"test_size": 500000
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.01,
"learning_rate": 0.1,
"max_depth": 8,
"reg_alpha": 1.0,
"reg_lambda": 0.1,
"reg_lambda": 1.0,
"n_estimators": 500
}
}
Expand All @@ -23,8 +31,8 @@
"data": {
"dataset": "medical_charges_nominal",
"split_kwargs": {
"train_size": 0.1,
"test_size": 0.9
"train_size": 0.5,
"test_size": 0.5
}
},
"algorithm": {
Expand All @@ -39,54 +47,76 @@
},
{
"data": {
"dataset": "year_prediction_msd",
"dataset": "nyc_taxi_green",
"preprocessing_kwargs": {
"category_encoding": ["ordinal", "ignore"]
},
"split_kwargs": {
"train_size": 0.25,
"test_size": 0.75
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.1,
"reg_alpha": 0.5,
"reg_lambda": 0.5,
"n_estimators": 200
"max_depth": 8,
"reg_lambda": 1.0,
"n_estimators": 500
}
}
},
{
"data": {
"dataset": "hepmass",
"dataset": "qsar_tid_11",
"split_kwargs": {
"train_size": 200000,
"test_size": 1000000
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.1,
"reg_alpha": 1.0,
"learning_rate": 0.05,
"max_depth": 6,
"subsample": 0.8,
"colsample_bytree": 0.8,
"reg_lambda": 1.0,
"max_leaves": 256,
"n_estimators": 500
"n_estimators": 1000
}
}
},
{
"data": {
"dataset": "superconductivity",
"split_kwargs": {
"test_size": 0.2,
"random_state": 42
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.02,
"max_depth": 7,
"subsample": 0.8,
"colsample_bytree": 0.8,
"reg_lambda": 1.0,
"n_estimators": 1500
}
}
},
{
"data": {
"dataset": "gisette",
"dataset": "year_prediction_msd",
"split_kwargs": {
"train_size": 2000,
"test_size": 5000
"train_size": 0.5,
"test_size": 0.5
}
},
"algorithm": {
"estimator_params": {
"learning_rate": 0.15,
"max_leaves": 256,
"colsample_bytree": 0.1,
"colsample_bynode": 0.1,
"n_estimators": 100
"learning_rate": 0.1,
"reg_alpha": 0.5,
"reg_lambda": 0.5,
"n_estimators": 200
}
}
}
Expand Down
18 changes: 18 additions & 0 deletions sklbench/benchmarks/common.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,9 +15,11 @@
# ===============================================================================

import argparse
import inspect
import json
from typing import Dict

from ..datasets.common import preprocess_x
from ..utils.bench_case import get_bench_case_value, get_data_name
from ..utils.custom_types import BenchCase
from ..utils.logger import logger
Expand Down Expand Up @@ -50,6 +52,22 @@ def enrich_result(result: Dict, bench_case: BenchCase) -> Dict:
f"`{result['library']}` to `sklearnex` in benchmark output."
)
result["library"] = "sklearnex"
preproc_kwargs = get_bench_case_value(bench_case, "data:preprocessing_kwargs", dict())
preproc_defaults = {
name: param.default
for name, param in inspect.signature(preprocess_x).parameters.items()
}
result.update(
{
"category_encoding": preproc_kwargs.get(
"category_encoding", preproc_defaults["category_encoding"]
),
"replace_nan": preproc_kwargs.get(
"replace_nan", preproc_defaults["replace_nan"]
),
"normalize": preproc_kwargs.get("normalize", preproc_defaults["normalize"]),
}
)
taskset = get_bench_case_value(bench_case, "bench:taskset", None)
if taskset is not None:
result.update({"taskset": taskset})
Expand Down
Loading
Loading