PolynomialDGP

PolynomialDGP generates datasets with polynomial interactions between features. The complexity parameter controls the interaction degree: low complexity uses main effects only, medium adds adjacent-pair cross-terms, and high introduces higher-order interactions.

Quick Start

import synthbench
from synthbench import BenchPipeline, PolynomialDGP

dgp = PolynomialDGP(complexity="medium", task_type="regression", random_state=0)
pipeline = BenchPipeline(dgp)
result = pipeline.run(n_samples=500, n_features=10, random_state=42)

print(result.X.shape)   # (500, 10)
print(result.y.shape)   # (500,)
print(list(result.metadata.keys()))

# Signal importances sum to 1.0
importances = result.metadata["signal_feature_importances"]
print(sum(importances.values()))  # 1.0

Parameters

Parameter	Default	Description
complexity	"medium"	Controls interaction degree: "low" = main effects, "medium" = cross-terms, "high" = higher-order
task_type	"regression"	"regression" for continuous target, "classification" for binary labels
random_state	0	Integer seed for reproducibility
class_weight	0.5	(Classification only) Fraction of samples in the positive class

Notes

• Feature importances are structural equal-weight importances (1/n_informative) rather than coefficient-squared, reflecting the polynomial structure of the ground truth.
• Medium complexity uses adjacent-pair cross-terms (feature i x feature i+1).
• Noise features receive exactly 0.0 importance in signal_feature_importances.