Slides at emitanaka.org/slides/DAF2021/edibble
edibble::start_design()
- Begin with
start_design()
library(edibble)start_design()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::start_design()
- Begin with
start_design()
library(edibble)start_design()
- This doesn't do much except create a new edibble design object.
- An edibble design contains an edibble graph (and later edibble table)
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::start_design()
- Begin with
start_design()
library(edibble)start_design()
- This doesn't do much except create a new edibble design object.
An edibble design contains an edibble graph (and later edibble table)
You can give it a name to your design — think of it as the title of your experiment
start_design("My diet experiment")
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_context()
- Set small notes that remind you the context of the experiment
start_design("My diet experiment") %>% set_context(aim = "Understand relation between diet and weight gain", experimenter = "Taylor Alwyn {.email tswift@fakemail.com}", "More details in {.file details.txt}")
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_context()
- Set small notes that remind you the context of the experiment
start_design("My diet experiment") %>% set_context(aim = "Understand relation between diet and weight gain", experimenter = "Taylor Alwyn {.email tswift@fakemail.com}", "More details in {.file details.txt}")
- Context data is preserved in the object, displayed when object is printed and can be exported
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_context()
- Set small notes that remind you the context of the experiment
start_design("My diet experiment") %>% set_context(aim = "Understand relation between diet and weight gain", experimenter = "Taylor Alwyn {.email tswift@fakemail.com}", "More details in {.file details.txt}")
- Context data is preserved in the object, displayed when object is printed and can be exported
Persistent reminder of experimental context so information is not lost in your email or elsewhere!
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 1
- A "unit" in edibble is any entity, physical or otherwise, that pertain to the experiment.
- A single integer is a shorthand for the number of levels.
start_design("My diet experiment") %>% set_units(subject = 20)
- Above specify there are 20 subjects for the experiment.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 1
- A "unit" in edibble is any entity, physical or otherwise, that pertain to the experiment.
- A single integer is a shorthand for the number of levels.
start_design("My diet experiment") %>% set_units(subject = 20)
- Above specify there are 20 subjects for the experiment.
- Under the hood, there is an edibble graph that contains:
- a node corresponding to the variable subject and
- 20 other nodes corresponding to the levels associated with subject.
Slides at emitanaka.org/slides/DAF2021/edibble
An edibble graph
The full graph:
set.seed(1)library(edibble)g <- start_design() %>% set_units(subject = 20)plot(g, view = "all")
# defaultplot(g, view = "high")
plot(g, view = "low", layout = igraph::layout_as_tree, asp = 10)
Slides at emitanaka.org/slides/DAF2021/edibble
An edibble table
start_design("My diet experiment") %>% set_units(subject = 20) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 2
- A string vector is a short hand for the names of the levels.
start_design("My diet experiment") %>% set_units(subject = c("Bettie", "Javid", "Yohan", "Marco", "Joani", "Tynika", "Lakendrick", "Stephanos", "Lavonda", "Benny", "Daniell", "Juanito", "Kele", "Delance", "Shekelia", "Meghan", "Lynzie", "Viraaj", "Jeffrey", "Sunni"))
Slides at emitanaka.org/slides/DAF2021/edibble
Seeing the level names in edibble table
start_design("My diet experiment") %>% set_units(subject = c("Bettie", "Javid", "Yohan", "Marco", "Joani", "Tynika", "Lakendrick", "Stephanos", "Lavonda", "Benny", "Daniell", "Juanito", "Kele", "Delance", "Shekelia", "Meghan", "Lynzie", "Viraaj", "Jeffrey", "Sunni")) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 3
- The argument name can be anything!
- So user may use names that match the experimental context.
start_design("My diet experiment") %>% set_units(fly = 20)
- Above reads that there are 20 flies in this diet experiment.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 3
- The argument name can be anything!
- So user may use names that match the experimental context.
start_design("My diet experiment") %>% set_units(fly = 20)
- Above reads that there are 20 flies in this diet experiment.
start_design("My diet experiment") %>% set_units(pig = 20)- Now it reads that there are 20 pigs in this diet experiment.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 4
- You can add "block" units, or any other types of "units".
start_design("My diet experiment") %>% set_units(pen = 10, pig = 50)
- Above reads that there are 10 pens and 50 pigs.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_units() Part 4
- You can add "block" units, or any other types of "units".
start_design("My diet experiment") %>% set_units(pen = 10, pig = 50)
Above reads that there are 10 pens and 50 pigs.
What do you think happens if we
serve_table()on this? What's your expectation of the output?
Slides at emitanaka.org/slides/DAF2021/edibble
Relationship between variables
We say that an edibble graph is reconcilable to an edibble table if for every variable, each level of the variable has a single linkage with level of another variable.
- Below edibble graph cannot be reconciled to an edibble table
start_design("My diet experiment") %>% set_units(pen = 10, pig = 50) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::nested_in() Part 1
- But the units must be related in someway.
start_design("My diet experiment") %>% set_units(pen = 10, pig = nested_in(pen, 5))
- Above reads that there are 10 pens and 5 pigs in each pen.
Slides at emitanaka.org/slides/DAF2021/edibble
The edibble table with nesting structure
start_design("My diet experiment") %>% set_units(pen = 10, pig = nested_in(pen, 5)) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::nested_in() Part 2
- More than one level of nesting:
start_design("My diet experiment") %>% set_units(pen = 10, pig = nested_in(pen, 5), time = nested_in(pig, 3)) %>% serve_table()
- Above reads that there are 10 pens, 5 pigs in each pen, and 3 time points observed for each pig.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::nested_in() Part 3
- Syntactic sugar for unbalanced unit structure. Again, the units don't need to be physical objects
start_design("My diet experiment") %>% set_units(pen = 10, pig = nested_in(pen, 1:2 ~ 5, 3 ~ 4, . ~ 2), time = nested_in(pig, 3))
- Above reads that there are 10 pens, 5 pigs in pens 1 & 2, 4 pigs in pen 3, and 2 pigs in the remaining pens (so a total of 2\times 5 + 4 + 2 \times 7 = 28 pigs), and 3 time points observed for each pig.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::nested_in() Part 4
- You can refer units by its label instead of level:
start_design("My diet experiment") %>% set_units(pen = c("A", "B", "C", "D", "E", "F", "G", "H", "I", "J"), pig = nested_in(pen, c("A", "B") ~ 5, "C" ~ 4, . ~ 2), time = nested_in(pig, 3))
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_trts() Part 1
- Treatment factors are like "units" but some distinguishable attributes added to the object if you use
set_trts.
start_design("My diet experiment") %>% set_trts(diet = c("NF", "HCD", "HFD", "HPD"))
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_trts() Part 1
- Treatment factors are like "units" but some distinguishable attributes added to the object if you use
set_trts.
start_design("My diet experiment") %>% set_trts(diet = c("NF", "HCD", "HFD", "HPD"))
- Below looks the same as above but also encodes another (long) label as well. Note: long label not used yet downstream.
start_design("My diet experiment") %>% set_trts(diet = c( "normal food" = "NF", "high-carbon diet" = "HCD", "high fat diet" = "HFD", "high protein diet" = "HPD"))Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_trts() Part 2
- You can set the treatment first then units or vice-versa.
start_design("My diet experiment") %>% set_units(pen = 10) %>% set_trts(diet = c("NF", "HCD", "HFD", "HPD"))
start_design("My diet experiment") %>% set_trts(diet = c("NF", "HCD", "HFD", "HPD")) %>% set_units(pen = 10)
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_trts() Part 3
- Factorial treatments
start_design("My diet experiment") %>% set_trts(type = c("carb", "fat", "protein"), level = c("high", "low"))
Slides at emitanaka.org/slides/DAF2021/edibble
Setting variables later
- You can break the unit or treatment factors to another function later in the pipeline:
start_design("My diet experiment") %>% set_trts(type = c("carb", "fat", "protein")) %>% set_units(pen = 10) %>% set_trts(level = c("high", "low")) %>% set_units(pig = nested_in(pen, 5)) %>% set_units(time = nested_in(pig, 3))
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::allocate_trts() Part 2
des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::allocate_trts() Part 2
des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% serve_table()
Do you notice anything about the treatment order?
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::allocate_trts() Part 2
des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% serve_table()
Do you notice anything about the treatment order?
Yup, it's systematically ordered!
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::randomise_trts()
set.seed(1)des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% randomise_trts() %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::make_classical
- Still want the "named" experimental design?
make_classical("crd", t = 5, n = 20)
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_rcrds()
des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% randomise_trts() %>% set_rcrds(pen = avg_temp, pig = c(diseased, inital_weight, final_weight)) %>% serve_table()
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::set_rcrds()
des %>% allocate_trts(diet ~ pen, breed ~ pig) %>% randomise_trts() %>% set_rcrds(pen = avg_temp, pig = c(diseased, inital_weight, final_weight)) %>% serve_table()
- The intention of which metric to capture on what is clearly specified.
- What is an observational unit is more obvious.
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::export_design()
export_design(df, file = "design.xlsx")
Slides at emitanaka.org/slides/DAF2021/edibble
edibble::export_design()
export_design(df, file = "design.xlsx")
Slides at emitanaka.org/slides/DAF2021/edibble
Visualising designs with ggplot2
library(ggplot2)df1 %>% # make it normal data frame as_data_frame() %>% # plot using ggplot ggplot(aes(pig, "1", fill = diet)) + geom_tile(color = "black")
Slides at emitanaka.org/slides/DAF2021/edibble
Visualising designs with ggplot2
library(ggplot2)df1 %>% # make it normal data frame as_data_frame() %>% # plot using ggplot ggplot(aes(pig, "1", fill = diet)) + geom_tile(color = "black")
- Slightly painful if you want to quickly visualise your design.
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 1
- Just
autoplotit!
library(deggust)autoplot(df1)
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 2
- It makes some decision for you of how to plot which can be customised in two ways:
- modify scale and theme like any ggplot objects!
- as arguments in the
autoplotfunction
autoplot(df1)
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 3
- It makes some decision for you of how to plot which can be customised in two ways:
- modify scale and theme like any ggplot objects!
- as arguments in the
autoplotfunction
autoplot(df1) + # ggplot2 functions below theme_void() + scale_fill_viridis_d(option = "A")
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 4
- It makes some decision for you of how to plot which can be customised in two ways:
- modify scale and theme like any ggplot objects!
- as arguments in the
autoplotfunction
autoplot(df1, shape = "hexagon", text = TRUE, aspect_ratio = 0.5)
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 5
autoplot(df1, image = "images/pig.png") + theme_void()
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 6
- Nested design
set.seed(2021)start_design() %>% set_units(pen = 10, pig = nested_in(pen, 5)) %>% set_trts(breed = c("A", "B", "C")) %>% allocate_trts(breed ~ pig) %>% randomise_trts() %>% serve_table() %>% autoplot()
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 7
- What changed here?
set.seed(2021)start_design() %>% set_units(pen = 10, pig = nested_in(pen, 5)) %>% set_trts(breed = c("A", "B", "C")) %>% allocate_trts(breed ~ pen) %>% randomise_trts() %>% serve_table() %>% autoplot()
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 8
- Factorial experiment
set.seed(2021)start_design() %>% set_units(pig = 40) %>% set_trts(breed = c("A", "B", "C"), feed = c("X", "Y", "Z")) %>% allocate_trts(breed:feed ~ pig) %>% randomise_trts() %>% serve_table() %>% autoplot()
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 9
- Focus on the treatment you want:
set.seed(2021)start_design() %>% set_units(pig = 40) %>% set_trts(breed = c("A", "B", "C"), feed = c("X", "Y", "Z")) %>% allocate_trts(breed:feed ~ pig) %>% randomise_trts() %>% serve_table() %>% autoplot(trts = "breed")
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 10
- Is your design too big to fit in the plot?
set.seed(2021)start_design() %>% set_units(pen = 100, pig = nested_in(pen, 10)) %>% set_trts(breed = c("A", "B", "C")) %>% allocate_trts(breed ~ pig) %>% randomise_trts() %>% serve_table() %>% autoplot()
Slides at emitanaka.org/slides/DAF2021/edibble
deggust::autoplot() Part 10
- Is your design too big to fit in the plot?
- Subset it!
set.seed(2021)start_design() %>% set_units(pen = 100, pig = nested_in(pen, 10)) %>% set_trts(breed = c("A", "B", "C")) %>% allocate_trts(breed ~ pig) %>% randomise_trts() %>% serve_table() %>% dplyr::filter(pen %in% c("pen1", "pen2", "pen3", "pen4")) %>% autoplot() + annotate("text", x = 10, y = 4.7, label = "x 25 more")
Slides at emitanaka.org/slides/DAF2021/edibble
edibbleGUI::app()
- This app is bound to change.
- The development of the app will always lag from
edibble.
Some plans:
- Show corresponding code
- Export design table
- Host app on the web
- Better integration with
edibble
Slides at emitanaka.org/slides/DAF2021/edibble
What edibble does
tidyversedoes well for processes in (B)edibbleaims to tackle (A)
