Regression Benchmarks

All models are evaluated under default settings from their respective libraries.

• No hyperparameter tuning.
• Linear models and KNN are scaled for transparency.
• Same preprocessing for all models:
• Unique encoding for categorical features
• No dataset-specific tricks
• Metrics reported are 3-Fold Cross Validation means.
• Latency is measured on CPU.

Metrics shown:

• RMSE (CV Mean)
• RMSE (Test)
• Single Inference P95 (ms)

---

Diamonds

Model	CV RMSE	Test RMSE	Single P95 (ms)
SmartKNN	892.21	860.76	0.19
KNN	933.65	897.88	0.71
RandomForest	1153.65	745.21	34.46
DecisionTree	1392.59	1164.86	0.18
LightGBM	1216.61	1191.34	1.23
XGBoost	1339.78	1238.60	0.53
Ridge	1287.62	1297.21	0.12
CatBoost	1276.41	1436.24	1.19

---

OnlineNewsPopularity

Model	CV RMSE	Test RMSE	Single P95 (ms)
Ridge	16013.18	10839.29	0.19
CatBoost	11676.77	10908.77	1.29
LightGBM	11759.63	11021.90	1.14
RandomForest	11909.29	11247.33	36.10
SmartKNN	12560.55	11879.40	0.26
KNN	12455.76	11944.82	2.90
XGBoost	12410.22	13315.14	0.56
DecisionTree	17126.61	22562.39	0.16

---

Allstate Claims Severity

Model	CV RMSE	Test RMSE	Single P95 (ms)
CatBoost	1918.54	1846.01	1.55
LightGBM	1936.83	1869.23	1.34
XGBoost	2325.40	1866.70	0.60
RandomForest	1989.76	1897.76	38.26
Ridge	2085.99	2026.17	0.23
SmartKNN	2227.13	2141.85	0.40
KNN	2292.37	2216.83	19.54
DecisionTree	2907.48	2704.40	0.15

---

Buzz in Social Media (Twitter)

Model	CV RMSE	Test RMSE	Single P95 (ms)
RandomForest	162.14	137.73	37.77
Ridge	166.66	156.38	0.16
SmartKNN	160.30	166.91	0.37
KNN	163.58	172.84	38.89
DecisionTree	252.82	196.36	0.17
LightGBM	187.52	209.01	1.20
CatBoost	228.11	246.56	1.52
XGBoost	286.47	265.39	0.58

---

California Housing

Model	CV RMSE	Test RMSE	Single P95 (ms)
CatBoost	46494.19	45589.21	1.28
XGBoost	48866.09	46888.93	0.58
LightGBM	48476.02	47651.31	1.05
RandomForest	50266.48	49046.64	36.87
SmartKNN	61998.65	61613.37	0.18
KNN	62183.02	61572.17	0.65
DecisionTree	70126.62	69915.73	0.16
Ridge	68646.24	70158.56	0.11

---

Wine Reviews

Model	CV RMSE	Test RMSE	Single P95 (ms)
CatBoost	2.2060	2.1289	1.72
LightGBM	2.2113	2.1472	1.04
RandomForest	2.2291	2.1558	35.51
XGBoost	2.2990	2.1578	0.51
SmartKNN	2.3285	2.3331	0.21
KNN	2.3759	2.3559	0.69
Ridge	2.4561	2.4344	0.19
DecisionTree	3.1205	2.8906	0.15

---

California Environmental Conditions

Model	CV RMSE	Test RMSE	Single P95 (ms)
RandomForest	0.1122	0.1035	34.35
XGBoost	0.1206	0.1104	0.56
CatBoost	0.1156	0.1104	1.16
SmartKNN	0.1175	0.1127	0.22
LightGBM	0.1175	0.1138	1.02
KNN	0.1203	0.1190	3.71
DecisionTree	0.1593	0.1485	0.15
Ridge	0.1517	0.1550	0.18

---