doe-helper
Design of Experiments Helper
Run fewer experiments. Find the right answer faster. Define your factors and responses, and doe-helper picks the optimal experiment plan, runs it, and tells you exactly which settings matter — with statistical confidence, not guesswork.
11 Experiment Designs
Full factorial, fractional factorial, Plackett-Burman, Box-Behnken, central composite, Latin hypercube, definitive screening, Taguchi, D-optimal, and two mixture designs. Pick by goal or let the tool recommend one.
Automated & Manual Testing
Generate ready-to-run Bash or Python scripts that execute each experimental run, collect results, and recover from failures automatically. Or record results by hand — both workflows are first-class.
Statistical Analysis
ANOVA with F-tests and p-values, main effects and interaction estimates, Pareto charts, normal probability plots, and effect contribution percentages — all generated from your results in one command.
Response Surface Modeling
Fit quadratic models to your data, visualize 3D response surfaces, and find predicted optima. See exactly where diminishing returns set in and where interactions dominate.
Multi-Objective Optimization
Balance competing goals — maximize throughput while minimizing cost, or improve quality without sacrificing speed. Desirability functions and weighted optimization find the best compromise.
Interactive HTML Reports
One command produces a self-contained report with embedded charts, effect tables, and design summaries. Share with your team — no software required to view it.
Power Analysis
Before you run a single test, know whether your design can detect the effect sizes you care about. Avoid wasting runs on underpowered experiments.
Design Augmentation
Add fold-over runs to break aliases, star points for curvature, or center points for pure error. Extend an existing design without starting over.
220+ Ready-Made Templates
From chemical reactors to Kubernetes tuning to bread baking — start with a real-world template, customize it, and run. Each includes a config, simulator, and documentation.
Error Recovery
Generated runner scripts handle per-run failures gracefully — log the error, skip the run, and continue. No lost progress from a single bad test.
CSV & Worksheet Export
Export your design matrix as a CSV or printable markdown worksheet for lab use. Import results from spreadsheets or hand-record them interactively at the command line.
Design Evaluation
Before running anything, check D-efficiency, A-efficiency, and G-efficiency scores. Know whether your design can estimate the effects you need with the precision you expect.