design-of-experiments

Identify optimal experimental design strategies for Design of Experiments in Python.

34|7|Updated Nov 29, 2025

Search Tags:#bayesian optimization #doe #design of experiments #box-behnken #ccd

Ready to run

npx skills add https://github.com/jkitchin/skillz --skill design-of-experiments

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: design-of-experiments
Source: https://github.com/jkitchin/skillz/tree/main/skills/scientific/design-of-experiments
Command: npx skills add https://github.com/jkitchin/skillz --skill design-of-experiments

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyDOE3, dexpy, pycse, scikit-optimize, statsmodels, scipy, scikit-learn, modAL, GPy, matplotlib, seaborn, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Design of Experiments (DOE) guidance for planning, executing, and analyzing experiments with classical designs, Bayesian optimization, model-driven designs, and active learning to maximize information per run.

Core Features & Use Cases

Interactive DOE guidance: Question-driven recommendations for batch vs sequential designs.
Classical DOE: Full/fractional factorials, CCD, Box-Behnken, and screening designs.
Bayesian & active learning: Gaussian-process-based optimization and adaptive sampling.

Quick Start

Ask for a DOE plan to screen 6 factors in 20 runs, using a fractional factorial design, with follow-up optimization via Bayesian options.

Frequently Asked Questions about design-of-experiments

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I choose between classical DOE designs and Bayesian optimization for my experiment?▼

Classical designs like factorials and Box-Behnken suit screening and fixed budgets; Bayesian optimization adapts sequentially to maximize information per run. Choose classical DOE for predetermined factor ranges and Bayesian when you can run experiments iteratively and refine objectives.

What's the best way to screen many factors in a limited number of experimental runs?▼

Fractional factorial designs efficiently screen factors in fewer runs than full factorials by confounding high-order interactions. This Skill recommends appropriate fractional designs based on your factor count and run budget.

Can I use Bayesian optimization with Gaussian processes for sequential experimental design?▼

Yes. Gaussian-process-based Bayesian optimization adaptively selects next experiments by balancing exploration and exploitation, reducing total runs needed. This Skill integrates scikit-optimize and GPy for this workflow.

How do I analyze results and decide whether to optimize, explore further, or stop experiments?▼

DOE analysis compares factor effects and model fit to guide next steps. This Skill provides decision-tree guidance to recommend screening, optimization, model discrimination, or robustness tasks based on your findings.

Does this approach support active learning or adaptive sampling strategies?▼

Yes. Active learning via modAL selects experiments that maximize information gain about uncertain regions. This Skill integrates active learning for model-driven designs when sequential experimentation is feasible.

What do I need to prepare before running a DOE workflow?▼

Define your experimental question, list factors and ranges, set run budget, and choose design goal—screening, optimization, or robustness. This Skill guides you through a question-driven flow to select the right design method.