CI fixes

Author: Alexsandruss

configs/spmd/kmeans.json

@@ -7,8 +7,8 @@
         "kmeans": {
             "SETS": [
                 "common kmeans parameters",
-				"sklearn kmeans parameters",
-				"kmeans datasets",
+                "sklearn kmeans parameters",
+                "kmeans datasets",
                 "sklearnex spmd implementation",
                 "spmd default parameters",
                 "spmd kmeans parameters"

configs/testing/azure-pipelines-ci.json

@@ -101,13 +101,14 @@
             },
             {
                 "algorithm": {
-                    "estimator": ["SVC", "SVR"]
+                    "estimator": ["SVC", "SVR"],
+                    "estimator_params": { "max_iter": 1000 }
                 }
             },
             {
                 "algorithm": {
                     "estimator": ["NuSVC", "NuSVR"],
-                    "estimator_params": { "nu": 0.1 }
+                    "estimator_params": { "max_iter": 1000, "nu": 0.1 }
                 }
             },
             {

sklbench/report/compatibility.py

@@ -23,6 +23,17 @@
 
 
 def transform_results_to_compatible(results: pd.DataFrame):
+    # sklearn and sklearnex compatibility
+    if (results["library"] == "sklearnex").any():
+        # delete extra columns related to sklearnex only
+        results.drop(
+            inplace=True,
+            errors="ignore",
+            columns=[
+                "max_bins",
+                "min_bin_size",
+            ],
+        )
     # cuML compatibility
     if (
         (results["library"] == "cuml")

sklbench/runner/arguments.py

@@ -128,7 +128,7 @@ def add_runner_arguments(parser: argparse.ArgumentParser) -> argparse.ArgumentPa
         "--prefetch-datasets",
         default=False,
         action="store_true",
-        help="[EXPERIMENTAL] Load named datasets in parallel before running benchmarks.",
+        help="Load all requested datasets in parallel before running benchmarks.",
     )
     # workflow control
     parser.add_argument(

test-configuration.yml

@@ -26,5 +26,5 @@ steps:
       python -m sklbench -l DEBUG -p algorithm:library=sklearnex algorithm:estimator=KMeans data:dataset=a9a
     displayName: CLI arguments run
   - script: |
-      python -m sklbench -l INFO -c configs/testing/azure-pipelines-ci.json --report
+      python -m sklbench -l INFO -c configs/testing/azure-pipelines-ci.json --prefetch-datasets --report --diff-cols library --compatibility-mode
     displayName: CI config run