design-of-experiments
Expert guidance for Design of Experiments (DOE) in Python - interactive goal-driven design selection, classical DOE (factorial, response surface, screening), Bayesian optimization with Gaussian processes, model-driven optimal designs, active learning, and sequential experimentation; includes pyDOE3, pycse, GPyOpt, scikit-optimize, statsmodels
npx skills add majiayu000/claude-skill-registry --skill design-of-experiments-jkitchin-skillz-2 --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Provides interactive guidance for Design of Experiments (DOE) in Python, offering goal-driven recommendations across classical DOE (factorial, response surface, screening), Bayesian optimization with Gaussian processes, model-driven optimal designs, and active learning. It helps users decide between sequential and batch strategies, screening and optimization, exploration and exploitation, and outputs recommendations, alternative options, estimated runs, library/tool suggestions, and starter code to generate designs and start experiments.
How it works
The skill starts by asking a set of primary questions to determine goals, sequencing capability, experiment cost, number of factors, prior data or models, and constraints. Based on user answers, it provides: a recommended approach with rationale, trade-off options, an estimated number of runs, suggested libraries/tools, and example code. It then guides the user to generate a design, visualize it, export it to CSV, run experiments, and analyze results using Classical, Bayesian, or Model-driven methods. It includes example code blocks for classical DOE (e.g., central composite design), Bayesian optimization (gp_minimize-like workflow), model-driven D-optimal designs, and active learning with GP surrogates. The workflow emphasizes answering questions, receiving a tailored recommendation, generating designs, executing experiments, and analyzing results to inform decisions.
When to use it
Use when planning DOE projects to decide between sequential vs batch experimentation, screening vs optimization vs model discrimination, and exploration vs exploitation. It applies to scenarios with varying cost per experiment, different numbers of factors, and whether there is a prior model or data. It provides guidance on which approach fits batch-only or sequential contexts and offers tool recommendations accordingly.
What it can touch
Tools mentioned include GPyOpt, scikit-optimize, BoTorch, Ax, pyDOE3, dexpy, pycse, statsmodels, modAL, scikit-learn, and general Python code examples. The skill references generating designs, exporting to CSV, and executing experiments, as well as analyzing results with ANOVA, regression, Bayesian updating, and model fitting.
Caveats
License is MIT. The skill presents design patterns and example code but does not guarantee specific experimental outcomes and relies on user-provided data and context. It may reference multiple external libraries; users should ensure compatibility with their environment and dependencies.
# Design of Experiments (DOE) - Interactive Expert ## Overview Master experimental design through **interactive, goal-driven guidance** that asks the right questions to recommend the best approach for your situation. This skill covers classical DOE, Bayesian optimization, model-driven designs, and active learning—helping you choose between batch and sequential strategies, screening and optimization, and exploration and exploitation. **Core value:** Don't start with a method—start with questions. Based on your goals, budget, and constraints, get personalized recommendations for experimental design strategies that maximize information gain per experiment. ## CRITICAL: Start with Questions, Not Methods **When a user mentions DOE or experimental design, ASK THESE QUESTIONS FIRST:** ### Primary Questions (Ask Before Recommending): 1. **"What is your primary goal?"** - **Screening**: Identify which factors matter (many factors → few important) - **Optimization**: Find best settings for known factors - **Exploration**: Understand the system/build model - **Model discrimination**: Choose between competing models - **Robustness**: Minimize sensitivity to noise 2. **"Can you run experiments
- Overview
- CRITICAL: Start with Questions, Not Methods
- Primary Questions (Ask Before Recommending):
- Based on Answers, Recommend:
- Quick Decision Tree
- When to Use Each Approach
- Classical DOE (Batch Experiments)
- Bayesian Optimization (Sequential)
- Model-Driven DOE (Optimal Designs)
- Active Learning (Sequential Model Building)
- Quick Reference Table
- Quick Start Examples
- Example 1: Classical Response Surface (Batch)
- Example 2: pycse Surface Response
Classical DOE pip install pyDOE3 # Factorial, RSM, LHS pip install dexpy # Modern DOE library pycse for integrated RSM pip install pycse Bayesian Optimization pip install scikit-optimize # skopt - easiest to use pip install GPyOpt # Comprehensive BO pip install ax-platform # Meta's adaptive experimentation pip install botorch torch # Advanced BO (requires PyTorch)
What does the design-of-experiments skill do?
Expert guidance for Design of Experiments (DOE) in Python - interactive goal-driven design selection, classical DOE (factorial, response surface, screening), Bayesian optimization with Gaussian processes, model-driven optimal designs, active learning, and sequential experimentation; includes pyDOE3, pycse, GPyOpt, scikit-optimize, statsmodels
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill design-of-experiments-jkitchin-skillz-2 --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.