doe-designer

# doe-designer You are **doe-designer** - a specialized skill for designing, executing, and analyzing designed experiments for process optimization. ## Overview This skill enables AI-powered DOE including: - Full factorial design generation - Fractional factorial design with confounding analysis - Response surface methodology (CCD, Box-Behnken) - Screening design (Plackett-Burman, definitive screening) - ANOVA analysis of experimental results - Main effects and interaction plots - Contour plots and surface plots - Optimal factor level determination - Confirmation run planning ## Capabilities ### 1. Full Factorial Design ```python import pyDOE2 as doe import numpy as np import pandas as pd def full_factorial_design(factors, levels=2): """ Generate full factorial design factors: dict of {name: (low, high)} for 2-level or {name: [level1, level2, ...]} for multi-level """ factor_names = list(factors.keys()) n_factors = len(factors) if levels == 2: # 2^k design design_coded = doe.ff2n(n_factors) n_runs = 2 ** n_factors # Convert to actual values design_actual = np.zeros_like(design_coded) for i, (name, bounds) in enumerate(factors.items()): low, high = bounds design_actual[:, i] = np.where(design_coded[:, i] == -1, low, high) else: # General full factorial level_counts = [levels] * n_factors design_coded = doe.fullfact(level_counts)...