README.Rmd

---
title: "Design of Experiments in R"
output: md_document
---

# Introduction

This repository is designed to be a refrence for myself and others for using R
in the design of experiments. It is a living document to store code snippets and
reminders of how to achieve outcomes given inputs and constraints.

This repository is heavily influenced by ["Statistical Methods for Mineral Engineers"](https://jktech.com.au/statistical-methods-mineral-engineers) by the
incredibly influential Prof. TJ Napier-Munn and the CRAN Task View ["Design of Experiments (DoE) & Analysis of Experimental Data"](https://cran.r-project.org/web/views/ExperimentalDesign.html) by Ulrike
Groemping.

## High Level Notes

* Use blocking to account for factors that probably have an effect but aren't interesting.
  - Date
  - Batch of materials
  - Person performing experiments
  - Equipment the test is performed on
* Pay attention to the [resolution](https://en.wikipedia.org/wiki/Fractional_factorial_design#Resolution) of the study. Stay within III-V most of the time. 
  - I  : Not Useful
  - II : Not useful, can't distinguish main effects
  - III: Useful, estimate main effects but can't distinguish two-factor interactions
  - IV : Useful, can estimate main and two-factor effects but two factor interactions can be confounded
  - V  : Useful, can estimate main, two-factor, two-factor interactions and three factor interactions
  - VI : Rarely useful, overkill, estimate all above factors unless there is other three-factor interactions


## Fractional Factorial Designs with 2-Level Factors

* Used When you have On-Off style interactions or High-Low effects.

Below is an example of a flotation test to measure bubble size given three
parameters and three replications. The experiment is of a resolution "III" and
will not be able to identify two-factor interactions.

```{r packages, results='asis', echo=TRUE}
pacman::p_load(tidyverse, FrF2)

example1 = FrF2(factor.names = c("frother","collector","salinity"),
                resolution = 3,
                replications = 3)
example1_runorder = attr(example1,"run.order")

example1 %>% bind_cols(example1_runorder) %>% knitr::kable(format = "markdown")

```