Legacy posts.
The original Experimental Design Hub blog. Earlier tutorials on Design of Experiments and Bayesian Optimization, kept here for reference. Newer writing lives in the journal.
- Bayesian Optimization
Understanding Bayesian Optimization
Discover how Bayesian Optimization revolutionizes experimental design by using machine learning to intelligently guide experiments, saving time and resources compared to traditional DoE methods.
- Bayesian Optimization
Testing Bayesian Optimization for Lab Experiments
Comparing Bayesian Optimization against random sampling and fractional factorial design using real lab data. Learn how BayBE achieves 98% success rate vs 50% for random approaches in optimizing coating hardness.
- Bayesian Optimization
A Technical Guide to Bayesian Optimization
How Bayesian Optimization decides which experiment to run next: surrogate models approximate your expensive experiments, and acquisition functions balance exploring the unknown with exploiting what looks promising.
- Bayesian Optimization · Python Examples
Getting Started with Bayesian Optimization (BayBE)
A practical walkthrough of using the BayBE Python package to optimize chemical processes efficiently
- Bayesian Optimization · Python Examples
Getting Started with Bayesian Optimization (NEXTorch)
A practical walkthrough of using the NEXTorch Python package to optimize chemical processes efficiently
- Bayesian Optimization · Python Examples
Getting Started with Bayesian Optimization (BoFire)
A practical walkthrough of using the BoFire Python package to optimize chemical processes efficiently
- Bayesian Optimization
How Many Experiments Do I Really Need?
Active learning, domain expertise, or random selection: which gets you the best predictions from the fewest experiments?
- DoE
What is Design of Experiments? (DoE Basics)
Learn the fundamentals of Design of Experiments (DoE) - a structured approach to experimental planning that maximizes information from minimal trials. Perfect introduction for chemical process optimization and experimental design.
- DoE
Key DoE Terms
Your essential cheat sheet for Design of Experiments terminology. Learn the key terms like factors, factor levels, response variables, and more to master DoE conversations and concepts.
- DoE
Principles of DoE: Randomization, Replication, Blocking
Learn the three fundamental principles of Design of Experiments that ensure high-quality, reproducible data: replication, randomization, and blocking. Essential DoE basics for reliable experimental results.
- DoE
Main Effects & Interaction Effects Explained
Understand main effects and interaction effects in factorial design. Learn how factors can work independently or synergistically in Design of Experiments through practical examples and visualization techniques.
- DoE
The Basics of Designing Better Experiments
Discover why one-factor-at-a-time (OFAT) approaches miss crucial interactions and how systematic design space coverage leads to full factorial designs. Learn to plan experiments that capture maximum information with minimal runs.
- DoE
A step by step example of a full factorial design
Learn what a full factorial design looks like in practice and how to create and analyze one—using a real filtration-rate dataset. We’ll read main-effects and interaction plots without heavy stats.
- DoE · Python Examples
Create a Full Factorial Design in Python
Step-by-step guide for beginners: install pyDOE3, create 2^k designs, add factor names, map to real units, randomize, and export ready-to-run experiment plans.
- DoE
Introducing Fractional & Central Composite Designs
When to move past full factorial design; how to use fractional designs to screen and central composite designs to optimize.
- DoE
Example of a Fractional Factorial Design
A step-by-step walkthrough of a fractional factorial design: why, when, and how to use it, with a practical example and visualizations.
- DoE · Python Examples
Create a Fractional Factorial Design in Python
Learn to create efficient fractional factorial designs using pyDOE3: reduce experiment costs while maintaining statistical power through strategic confounding and resolution optimization.
- DoE
Mathematical Models in DOE
Learn when and how to build mathematical models from your DOE results. Turn your experimental insights into predictive equations for complex optimization goals.
- DoE
How to perform ANOVA
Learn how to properly build models in DoE using ANOVA. Discover the systematic approach to determine which parameters truly matter and build reliable predictive models.
- DoE
Understanding the ANOVA Table Output
A practical guide to reading and interpreting ANOVA tables in Design of Experiments. Learn what each column means and how to use the output for decision-making.
- DoE
Testing ANOVA Assumptions
Learn to validate your ANOVA results by checking critical assumptions. Discover what can go wrong and how to spot problems before they invalidate your conclusions.
- DoE
QQ-Plots Explained
Learn how to use Quantile-Quantile plots to validate ANOVA assumptions and ensure your statistical models are reliable. Discover what QQ-plots reveal about data normality and how to interpret different patterns.
- DoE
Residual Analysis
Learn how to validate your DoE models through residual analysis, including visual comparisons, R-squared calculations, and residual plots to ensure your experimental conclusions are reliable.
- DoE · Python Examples
How to Perform ANOVA with Python
Learn to perform ANOVA analysis using Python and statsmodels. From basic model fitting to advanced techniques like backward elimination and quadratic terms.
- DoE
A step by step example of a Central Composite Design (CCD)
Learn how to set up and analyze a Central Composite Design through a practical example. Discover how CCD helps find optimal conditions while detecting curvature in your process.
- DoE · Python Examples
Create a Central Composite Design in Python
Practical, reproducible steps to build a CCD with Python.
- DoE
Central Composite Design vs. Box-Behnken Design
Learn the key differences between Central Composite and Box-Behnken designs, and discover which response surface design is best for your optimization experiments.
- DoE · Python Examples
How to create a Box-Behnken Design in Python
Learn to create efficient Box-Behnken designs using pyDOE3: a practical guide for optimizing processes with fewer experiments while exploring quadratic effects and curvature.
- DoE · Python Examples
A full factorial design in Python from beginning to end
A comprehensive step-by-step guide to implementing 2-level full factorial designs using Python. Learn to create experimental plans, visualize main effects and interactions, perform ANOVA, conduct model diagnostics, and generate 3D surface plots - all with practical code examples and real filtration rate data.
- DoE
Understanding Systematic and Random Errors
Learn to identify, understand, and manage systematic and random errors in experimental design. Discover how these errors impact accuracy and precision, and master the strategies to minimize their effects on your DoE results.
- DoE
Why Blocking Matters
A real-world example showing what happens when you skip blocking in your DoE. Learn from this epoxy coating experiment how batch-to-batch variability can confound your results—and how blocking prevents it.
- DoE
Response Surface Methodology (RSM)
Learn how Response Surface Methodology uses sequential experiments and polynomial models to navigate the optimization landscape and find better process conditions step by step.
- DoE
The Path of Steepest Ascent
Learn how to use the path of steepest ascent to optimize processes systematically. Includes practical examples for linear models with and without interactions.
- DoE · Python Examples
Why You Should Always Code Your Variables
Learn why coding variables is essential in DoE. Discover how standardizing factor scales prevents misleading coefficients and enables accurate model interpretation.
- DoE · Python Examples
Blocking in Practice: How to Actually Do It
Learn how blocking is implemented in experimental design through confounding, see a practical comparison of blocked vs. unblocked designs, and understand why randomization alone isn't enough.
- DoE
Active Effects: Not Everything Matters Equally
Learn what active effects are, how to distinguish them from inactive ones, and why the principles of effect sparsity and effect hierarchy are the foundation of efficient experimental design.
- DoE
Foldover Designs
Learn how foldover designs resolve confounding in fractional factorial experiments by adding a mirror-image set of runs, with a practical example showing how a misleading effect gets unmasked.
- DoE
All-Subset Regression
Instead of removing terms one by one, all-subset regression evaluates every possible combination of factors and picks the best model directly. Here's how it works and how it compares to ANOVA.