In-Silico DOE Lab
A reproducible biostatistics and DOE virtual lab rebuilt from the source Factory Excel 1–6 generators.
These are synthetic data for learning
They are not patient, process, or experimental evidence. The same settings, seed, and engine version reproduce the same result.
Choose a lab
Single-response DOE lab
Separate main effects from interactions for one response.
Experiment settings
Relationship to source
The actual formulas and coefficients from the user-authored Factory workbooks are preserved. Only volatile RAND/NORMINV was replaced by a versioned seeded generator.
The actual formulas and coefficients from the user-authored Factory workbooks are preserved. Only volatile RAND/NORMINV was replaced by a versioned seeded generator.
Data-generating formula
Y1 = 65 + 2·A − 9.6·B + 0.5·A×B + 4·C − 1.1·A×C + 3·B×C + ε, ε ~ N(0, 1²)Factor ranges
| Factor | Low | Center | High | Unit |
|---|---|---|---|---|
| Factor A active | 260 | 280 | 300 | a.u. |
| Factor B active | 50 | 75 | 100 | a.u. |
| Factor C active | 40 | 50 | 60 | a.u. |
| Factor D | 12 | 15 | 18 | a.u. |
| Factor E | 1 | 1.5 | 2 | a.u. |
| Factor F | 180 | 200 | 220 | a.u. |
| Factor G | 30 | 37 | 44 | a.u. |
Generated results
Run order and response
True effects inside the generator: Y1
These bars reveal the data-generating truth. In real research it is unknown and must be estimated from analysis.
Analysis-ready data preview
| Run | Unit | A (Actual) | B (Actual) | C (Actual) | Y1 |
|---|---|---|---|---|---|
| 1 | EU-01-01 | 300 | 100 | 60 | 63.679 |
| 1 | EU-01-02 | 300 | 100 | 60 | 64.01 |
| 1 | EU-01-03 | 300 | 100 | 60 | 64.588 |
| 2 | EU-02-01 | 300 | 50 | 60 | 74.673 |
| 2 | EU-02-02 | 300 | 50 | 60 | 77.032 |
| 2 | EU-02-03 | 300 | 50 | 60 | 76.697 |
| 3 | EU-03-01 | 280 | 75 | 50 | 62.84 |
| 3 | EU-03-02 | 280 | 75 | 50 | 64.124 |
| 3 | EU-03-03 | 280 | 75 | 50 | 63.367 |
| 4 | EU-04-01 | 260 | 50 | 60 | 75.462 |
| 4 | EU-04-02 | 260 | 50 | 60 | 75.575 |
| 4 | EU-04-03 | 260 | 50 | 60 | 76.548 |
| 5 | EU-05-01 | 300 | 50 | 40 | 77.45 |
| 5 | EU-05-02 | 300 | 50 | 40 | 77.609 |
| 5 | EU-05-03 | 300 | 50 | 40 | 77.164 |
| 6 | EU-06-01 | 260 | 100 | 60 | 60.363 |
| 6 | EU-06-02 | 260 | 100 | 60 | 59.044 |
| 6 | EU-06-03 | 260 | 100 | 60 | 60.797 |
| 7 | EU-07-01 | 300 | 100 | 40 | 53.592 |
| 7 | EU-07-02 | 300 | 100 | 40 | 51.289 |
| 7 | EU-07-03 | 300 | 100 | 40 | 52.148 |
| 8 | EU-08-01 | 260 | 50 | 40 | 70.488 |
| 8 | EU-08-02 | 260 | 50 | 40 | 70.251 |
| 8 | EU-08-03 | 260 | 50 | 40 | 70.051 |
| 9 | EU-09-01 | 280 | 75 | 50 | 65.45 |
| 9 | EU-09-02 | 280 | 75 | 50 | 66.093 |
| 9 | EU-09-03 | 280 | 75 | 50 | 65.006 |
| 10 | EU-10-01 | 260 | 100 | 40 | 45.844 |
| 10 | EU-10-02 | 260 | 100 | 40 | 43.3 |
| 10 | EU-10-03 | 260 | 100 | 40 | 44.78 |
Continue in JMP or Minitab
Download the analysis CSV and open it in your software. This lab focuses on what information a design creates and what statistical ideas a graph reveals—not a sequence of menu clicks.
This tool does not automatically pass or fail an experiment. Interpret plots, replicate structure, variation, and research context together.
Tool Guide
Definition
An educational virtual laboratory that reproduces the user-authored Factory Excel generators on the web. It supports DOE and biostatistics practice without exposing real research data. Source main effects, interactions, quadratic terms, and noise structures are preserved; only volatile RAND/NORMINV is replaced by a versioned seeded generator.
Purpose
Inspect spread, extremes, and replicate structure alongside the mean; compare main effects, interactions, and curvature; and practice a sequential path from screening to response surfaces. Create reproducible JMP or Minitab CSV files without confidential data.
How to Use
① Choose a lab and a source-labeled preset. ② Set design, seed, independent replicates, and technical replicates. ③ Run the experiment and read the plots and summaries. ④ Download analysis CSV, raw CSV, dictionary, and receipt. Analysis rows are technical-replicate means; identical settings, seed, and versions reproduce the table.
Examples
Example: a single-response full-factorial lab with 3 independent and 2 technical replicates creates 11×3=33 analysis rows and 66 raw rows. Set variation to zero to reveal the deterministic source signal. Generated data are educational; the tool does not automatically declare pass/fail or an optimum.