experimental-design

Name: experimental-design
Author: K-Dense-AI

296installs3.1kforks

How to install experimental-design?

k-dense-ai/scientific-agent-skills/experimental-design

$npx -y skills add K-Dense-AI/scientific-agent-skills --skill experimental-design

Installs into the current project.

Summary

Design experiments and studies BEFORE data is collected — choosing a design, randomizing, blocking, and laying out treatment combinations so the results will actually be interpretable. Use whenever someone is planning a study, asks how to assign subjects/samples to groups, mentio

Files · 1

View on GitHub

SKILL.md

1	# Experimental Design
2
3	## Overview
4
5	The design of a study — how units are assigned to conditions, what is held constant, what is varied, and in what structure — determines what questions the data can answer. No analysis can rescue a confounded or pseudoreplicated design after the fact. This skill is about the decisions made before data collection: picking a design that isolates the effect of interest, randomizing to license causal claims, blocking to remove known nuisance variation, and structuring multi-factor experiments so effects are estimable rather than tangled together.
6
7	The three ideas behind almost every good design (Fisher's principles):
8	- Randomization — assign treatments at random so that confounders, known and unknown, are balanced in expectation. This is what turns a comparison into a causal claim.
9	- Replication — independent repetition at the right level, so you can estimate variability and your effects aren't artifacts of a single unit. The most common fatal error is pseudoreplication: counting repeated measurements on the same unit as independent replicates.
10	- Blocking / local control — group similar units (by batch, day, site, litter) and randomize within blocks, removing that nuisance variation from the error term instead of letting it inflate noise.
11
12	This skill helps you choose among design types, generate the actual randomization or DOE layout (with reproducible scripts), and avoid the structural mistakes that make data uninterpretable.
13
14	## When to Use This Skill
15
16	- Planning any comparative experiment or trial and deciding how to assign units
17	- Randomizing subjects/samples to arms (simple, blocked, stratified, or cluster)
18	- Removing nuisance variation by blocking or stratification
19	- Designing multi-factor experiments: full or fractional factorial, screening designs
20	- Optimizing a response over continuous factors (response-surface designs)
21	- Within-subject / repeated-measures, crossover, split-plot, or Latin-square designs
22	- Cluster- or group-randomized designs (sites, clinics, classrooms, litters)
23	- Deciding the number and level of replicates and avoiding pseudoreplication
24	- Sequential, group-sequential, or adaptive designs with interim analyses
25	- Laying out plates/batches and randomizing run order to defeat drift
26
27	## Installation
28
29	```bash
30	uv pip install "numpy>=1.26" "pandas>=2.0" pyDOE3
31	```
32
33	`pyDOE3` is the maintained successor to pyDOE/pyDOE2 and supplies factorial,
34	fractional-factorial, Plackett-Burman, central-composite, Box-Behnken, and
35	Latin-hypercube generators. The bundled scripts wrap it to return designs in real
36	factor units with named columns and randomized run order.
37
38	---
39
40	## Choosing a design
41
42	Start from the question and the structure of your units, not from a favorite design.
43
44	```
45	What are you trying to learn?
46	│
47	├─ Compare a few predefined conditions (A vs B vs C)?
48	│ ├─ Units independent, possibly with a known nuisance factor (day, batch, site)?
49	│ │ → Completely randomized (no nuisance) or RANDOMIZED BLOCK design.
50	│ ├─ Each unit can receive every condition in sequence (washout possible)?
51	│ │ → CROSSOVER / repeated-measures design (more power, watch carry-over).
52	│ └─ You can only randomize groups, not individuals (schools, clinics)?
53	│ → CLUSTER-randomized design (analyze at the cluster level; see pseudoreplication).
54	│
55	├─ Screen MANY factors (5+) to find the few that matter?
56	│ → FRACTIONAL FACTORIAL or PLACKETT-BURMAN screening design.
57	│
58	├─ Quantify main effects AND interactions among a handful of factors?
59	│ → FULL 2^k FACTORIAL design.
60	│
61	├─ Find the settings that OPTIMIZE a response (curvature matters)?
62	│ → RESPONSE-

People who liked this also viewed

6 picks

Type

experimental-design

296installs3.1kforks

View on GitHub

How to install experimental-design?

k-dense-ai/scientific-agent-skills/experimental-design

$npx -y skills add K-Dense-AI/scientific-agent-skills --skill experimental-design

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Experimental Design
2
3	## Overview
4
5	The design of a study — how units are assigned to conditions, what is held constant, what is varied, and in what structure — determines what questions the data can answer. No analysis can rescue a confounded or pseudoreplicated design after the fact. This skill is about the decisions made before data collection: picking a design that isolates the effect of interest, randomizing to license causal claims, blocking to remove known nuisance variation, and structuring multi-factor experiments so effects are estimable rather than tangled together.
6
7	The three ideas behind almost every good design (Fisher's principles):
8	- Randomization — assign treatments at random so that confounders, known and unknown, are balanced in expectation. This is what turns a comparison into a causal claim.
9	- Replication — independent repetition at the right level, so you can estimate variability and your effects aren't artifacts of a single unit. The most common fatal error is pseudoreplication: counting repeated measurements on the same unit as independent replicates.
10	- Blocking / local control — group similar units (by batch, day, site, litter) and randomize within blocks, removing that nuisance variation from the error term instead of letting it inflate noise.
11
12	This skill helps you choose among design types, generate the actual randomization or DOE layout (with reproducible scripts), and avoid the structural mistakes that make data uninterpretable.
13
14	## When to Use This Skill
15
16	- Planning any comparative experiment or trial and deciding how to assign units
17	- Randomizing subjects/samples to arms (simple, blocked, stratified, or cluster)
18	- Removing nuisance variation by blocking or stratification
19	- Designing multi-factor experiments: full or fractional factorial, screening designs
20	- Optimizing a response over continuous factors (response-surface designs)
21	- Within-subject / repeated-measures, crossover, split-plot, or Latin-square designs
22	- Cluster- or group-randomized designs (sites, clinics, classrooms, litters)
23	- Deciding the number and level of replicates and avoiding pseudoreplication
24	- Sequential, group-sequential, or adaptive designs with interim analyses
25	- Laying out plates/batches and randomizing run order to defeat drift
26
27	## Installation
28
29	```bash
30	uv pip install "numpy>=1.26" "pandas>=2.0" pyDOE3
31	```
32
33	`pyDOE3` is the maintained successor to pyDOE/pyDOE2 and supplies factorial,
34	fractional-factorial, Plackett-Burman, central-composite, Box-Behnken, and
35	Latin-hypercube generators. The bundled scripts wrap it to return designs in real
36	factor units with named columns and randomized run order.
37
38	---
39
40	## Choosing a design
41
42	Start from the question and the structure of your units, not from a favorite design.
43
44	```
45	What are you trying to learn?
46	│
47	├─ Compare a few predefined conditions (A vs B vs C)?
48	│ ├─ Units independent, possibly with a known nuisance factor (day, batch, site)?
49	│ │ → Completely randomized (no nuisance) or RANDOMIZED BLOCK design.
50	│ ├─ Each unit can receive every condition in sequence (washout possible)?
51	│ │ → CROSSOVER / repeated-measures design (more power, watch carry-over).
52	│ └─ You can only randomize groups, not individuals (schools, clinics)?
53	│ → CLUSTER-randomized design (analyze at the cluster level; see pseudoreplication).
54	│
55	├─ Screen MANY factors (5+) to find the few that matter?
56	│ → FRACTIONAL FACTORIAL or PLACKETT-BURMAN screening design.
57	│
58	├─ Quantify main effects AND interactions among a handful of factors?
59	│ → FULL 2^k FACTORIAL design.
60	│
61	├─ Find the settings that OPTIMIZE a response (curvature matters)?
62	│ → RESPONSE-

People who liked this also viewed

6 picks

Type