Automated visualisation of experimental designs

👩🏻‍💻 Emi Tanaka @ Monash University

emi.tanaka@monash.edu
@statsgen
github.com/emitanaka
emitanaka.org

29th Novemebr 2022 Australasian Applied Statistics Conference 2022

`edibble` R package 📦

🎯 To construct an experimental design table (or tibble)

The Grammar of Experimental Designs (WIP)
emitanaka.org/edibble-book

An object oriented programming system to construct an experimental design by encapsulating the experimental structure and context in a cognitive approach.

Combine words (functions) = Unlimited sentences (designs)
Health check
Cooking experimental designs based on raw ingredients, rather than ready-made meals

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015")

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3))

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3)) %>% 
  set_trts(plant_diversity = c(1, 4, 16),
           temperature = c("0°C", "1.5°C", "3°C"))

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3)) %>% 
  set_trts(plant_diversity = c(1, 4, 16),
           temperature = c("0°C", "1.5°C", "3°C")) %>% 
  allot_trts(plant_diversity ~ plot,
                 temperature ~ subplot)

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3)) %>% 
  set_trts(plant_diversity = c(1, 4, 16),
           temperature = c("0°C", "1.5°C", "3°C")) %>% 
  allot_trts(plant_diversity ~ plot,
                 temperature ~ subplot) %>% 
  assign_trts("random")

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3)) %>% 
  set_trts(plant_diversity = c(1, 4, 16),
           temperature = c("0°C", "1.5°C", "3°C")) %>% 
  allot_trts(plant_diversity ~ plot,
                 temperature ~ subplot) %>% 
  assign_trts("random") %>% 
  serve_table()

Demonstration for `edibble` 📦

Steinauer et al. (2015) Plant diversity effects on soil microbial functions and enzymes are stronger than warming in a grassland experiment. Ecology. 96 (1) 99-112

library(edibble)
design("Steinauer et al. 2015") %>% 
  set_units(plot = 27,
            subplot = nested_in(plot, 3)) %>% 
  set_trts(plant_diversity = c(1, 4, 16),
           temperature = c("0°C", "1.5°C", "3°C")) %>% 
  allot_trts(plant_diversity ~ plot,
                 temperature ~ subplot) %>% 
  assign_trts("random") %>% 
  serve_table() -> des

Let’s save this edibble design object as des.

Internal structure of an `edibble` design object

Factor graph

plot(des) # OR plot(des, which = "factors") OR plot_fct_graph(des)

Internal structure of an `edibble` design object

Level graph

plot(des, which = "levels") # OR plot_lvl_graph(des)

`deggust` R package 📦

🎯 The design of experiment as a ggplot object

degust (verb) to taste or savor carefully or appreciatively.
deggust (package) to visualise edibble design objects appreciatively.

An edibble design object contains encoding with experimental structure already.
The deggust package aims to leverage this to quickly view this.

Easy-and-quick visualisation

library(deggust)
autoplot(des)

Yup that’s all!

Cutomise visualisation

library(deggust)
autoplot(des, aspect_ratio = 3/2)

Cutomise visualisation with `ggplot2`

library(deggust)
autoplot(des, aspect_ratio = 3/2) +
  theme(
    legend.position = "bottom",
    text = element_text(size = 18),
    plot.background = 
      element_rect(color = "black",
                   linetype = "dashed",
                   fill = "wheat"),
    plot.margin = margin(20, 20, 20, 20),
    plot.subtitle =
      element_text(margin = margin(0, 0, 10, 0))) +
  
  scale_fill_brewer(palette = "Reds")

Leveraging the power of established graphical system.

Recipe designs

Completely randomised designs

crd <- design("CRD") %>%
  set_units(unit = 30) %>%
  set_trts(trt = 5) %>%
  allot_trts(trt ~ unit) %>%
  assign_trts("random", seed = 2022) %>%
  serve_table()

autoplot(crd)

Randomised complete block designs

rcbd <- design("RCBD") %>%
  set_units(block = 10,
            unit = nested_in(block, 6)) %>%
  set_trts(trt = 6) %>%
  allot_trts(trt ~ unit) %>%
  assign_trts("random", seed = 2022) %>%
  serve_table()

autoplot(rcbd)

Modified block design

mbd <- design("Modified Block Design") %>%
  set_units(block = 10,
            unit = nested_in(block, 
                               1 ~ 3,
                               2 ~ 4,
                               3 ~ 5,
                               . ~ 6)) %>%
  set_trts(trt = 6) %>%
  allot_trts(trt ~ unit) %>%
  assign_trts("random", seed = 2022) %>%
  serve_table()

autoplot(mbd)

Graeco-Latin square design

glsd <- design("GLSD") %>%
  set_units(row = 9,
            col = 9,
            unit = crossed_by(row, col)) %>%
  set_trts(trt1 = 9,
           trt2 = 9) %>%
  allot_trts(trt1 ~ unit,
             trt2 ~ unit) %>%
  assign_trts("random", seed = 2022) %>%
  serve_table()

autoplot(glsd)

Hyper-Graeco-Latin Square Design

hglsd <- design("HGLSD") %>%
  set_units(block1 = 4,
            block2 = 4,
            block3 = 4,
            block4 = 4,
            unit = crossed_by(block1,
                              block2, 
                              block3,
                              block4)) %>%
  set_trts(trt = 4) %>%
  allot_trts(trt ~ unit) %>%
  assign_trts("random", seed = 2022) %>%
  serve_table()

autoplot(hglsd)

edibble R package

To construct experimental design objects.

On CRAN On GitHub github.com/emitanaka/edibble

deggust R package

To visualise edibble designs.

On GitHub
github.com/emitanaka/deggust

Slides at emitanaka.org/slides/AASC2022/

Feedback/comments/questions/requests/collaborations welcome!

emi.tanaka@monash.edu @statsgen @emitanaka@fosstodon.org github.com/emitanaka