Work on week 9

header.tex

@@ -1,3 +1,7 @@
+% Unicode
+\usepackage{pmboxdraw}
+
+
 \makeatletter
 \def\input@path{{..}}
 \makeatother

week9/slides.qmd

@@ -33,8 +33,9 @@ source(here::here("course_info.R"))
 
 # Assignments
 
-## Assignments 3 and 4
+## Assignments
 
+* Assignment 2 feedback
 * Assignment 3 due 10 May
 * Assignment 4 due 24 May
 
@@ -51,7 +52,7 @@ source(here::here("course_info.R"))
 * Heavier reliance on pandoc Lua filters
 * Uses pandoc templates for extensions
 
-\centerline{\includegraphics[width = 10cm]{qmd.png}}
+\centerline{\includegraphics[width = 10cm]{images/qmd.png}}
 
 ## Choose your engine
 
@@ -105,6 +106,8 @@ mtcars |>
 ```
 ````
 
+Reference the figure using `@fig-chunklabel`.
+
 ## Chunk options
 
 * Quarto consistently uses hyphenated options (`fig-width` rather than `fig.width`)
@@ -121,16 +124,227 @@ mtcars |>
 \fontsize{13}{17}\sf
 
 * Quarto extensions modify and extend functionality.
+* They are stored locally, in the `_extensions` folder alongside the qmd document.
 * See <https://quarto.org/docs/extensions/> for a list.
 * Templates are extensions used to define new output formats.
 * Journal templates at\newline <https://quarto.org/docs/extensions/listing-journals.html>
 * Monash templates at\newline <https://robjhyndman.com/hyndsight/quarto_templates.html>
 
+## quarto on the command line
+\fontsize{14}{14.5}\sf\vspace*{-0.4cm}
+
+* `quarto render` to render a quarto or Rmarkdown document.
+* `quarto preview` to preview a quarto or Rmarkdown document.
+* `quarto add <gh-org>/<gh-repo>` to add an extension from a github repository.
+* `quarto update <gh-org>/<gh-repo>` to update an extension
+* `quarto remove <gh-org>/<gh-repo>` to remove an extension
+* `quarto list extensions installed`
+* `quarto use template <gh-org>/<gh-repo>` to use existing repo as starter template.
+
+## Add a custom format
+
+From the CLI:\qquad  `quarto add numbats/monash-quarto-memo`\pause
+
+New folder/files added
+
+```{verbatim}
+├── _extensions
+│   └── numbats
+│       └── memo
+│           └── ...
+```
+
+\pause
+
+Update YAML
+
+```{verbatim}
+---
+title: "My new file using the `memo-pdf` format"
+format: memo-pdf
+---
+```
+
 ## Exercise
 
-* Create a quarto document using an html format
-* Add a code chunk to generate a figure with a figure caption.
+* Set up a new project.
+* Create a quarto document using an html format.
+* Add a code chunk to generate a figure with a caption.
 * Reference the figure in the text using `@fig-chunklabel`.
 * Add the monash memo extension and generate a pdf output.
 
 # targets
+
+## targets: reproducible computation at scale
+
+\placefig{0.5}{1.8}{width=5cm}{images/logo.png}
+
+\begin{textblock}{15}(0.5,8.5)
+\textcolor{gray}{\footnotesize Some images from https://wlandau.github.io/targets-tutorial}
+\end{textblock}
+
+\begin{textblock}{10}(6, 2)
+\begin{itemize}
+\item Supports a clean, modular, function-oriented programming style.
+\item Learns how your pipeline fits together.
+\item Runs only the necessary computation.
+\item Abstracts files as R objects.
+\item Similar to Makefiles, but with R functions.
+\end{itemize}
+\end{textblock}
+
+## Interconnected tasks
+
+\only<1>{\placefig{0.5}{2}{width=13cm}{images/workflow.png}}
+\only<2>{\placefig{0.5}{2}{width=13cm}{images/change.png}}
+\only<3>{\placefig{0.5}{2}{width=13cm}{images/downstream.png}}
+
+## Dilemma: short runtimes or reproducible results?
+
+\fullheight{images/decisions.png}
+
+## Let a pipeline tool do the work
+
+\fullwidth{images/pipeline_graph.png}\vspace*{-0.15cm}
+
+* Save time while ensuring computational reproducibility.
+* Automatically skip tasks that are already up to date.
+
+## Typical project structure
+
+```{verbatim}
+_targets.R # Required top-level configuration file.
+R/
+└── functions.R
+data/
+└── my_data.csv
+```
+
+### _targets.R
+\vspace*{-0.26cm}
+
+```{r}
+#| eval: false
+library(targets)
+tar_source() # source all files in R folder
+tar_option_set(packages = c("tidyverse", "fable"))
+list(
+  tar_target(my_file, "data/my_data.csv", format = "file"),
+  tar_target(my_data, read_csv(my_file)),
+  tar_target(my_model, model_function(my_data))
+)
+```
+
+## Generate `_targets.R` in working directory
+
+```{r}
+#| eval: false
+library(targets)
+tar_script()
+```
+
+
+## Useful targets commands
+
+* `tar_make()` to run the pipeline.
+* `tar_make(starts_with("fig"))` to run only targets starting with "fig".
+* `tar_read(object)` to read a target.
+* `tar_load(object)` to load a target.
+* `tar_load_everything()` to load all targets.
+* `tar_manifest()` to list all targets
+* `tar_visnetwork()` to visualize the pipeline.
+* `tar_destroy()` to remove all targets.
+* `tar_outdated()` to list outdated targets.
+
+## Debugging
+
+Errored targets to return `NULL` so pipeline continues.
+
+```{r}
+#| eval: false
+tar_option_set(error = "null")
+```
+
+\pause
+
+See error messages for all targets.
+
+```{r}
+#| eval: false
+tar_meta(fields = error, complete_only = TRUE)
+```
+
+\pause
+
+See warning messages for all targets.
+
+```{r}
+#| eval: false
+tar_meta(fields = warnings, complete_only = TRUE)
+```
+
+## Debugging
+\fontsize{14}{15.5}\sf
+
+* Try loading all available targets: `tar_load_everything()`. Then run the command of the errored target in the console.
+
+* Pause the pipeline with `browser()`
+
+* Use the debug option: `tar_option_set(debug = "target_name")`
+
+* Save the workspaces:
+
+  - `tar_option_set(workspace_on_error = TRUE)`
+  - `tar_workspaces()`
+  - `tar_workspace(target_name)`
+
+
+## Random numbers
+
+* Each target runs with its own seed based on its name and the global seed from `tar_option_set(seed = ???)`
+* So running only some targets, or running them in a different order, will not change the results.
+
+## Folder structure
+
+```{verbatim}
+├── .git/
+├── .Rprofile
+├── .Renviron
+├── renv/
+├── index.Rmd
+├── _targets/
+├── _targets.R
+├── _targets.yaml
+├── R/
+├──── functions_data.R
+├──── functions_analysis.R
+├──── functions_visualization.R
+├── data/
+└──── input_data.csv
+```
+
+## targets with quarto
+
+```{r}
+#| eval: false
+library(targets)
+library(tarchetypes)                                                 # <1>
+tar_source() # source all files in R folder
+tar_option_set(packages = c("tidyverse", "fable"))
+list(
+  tar_target(my_file, "data/my_data.csv", format = "file"),
+  tar_target(my_data, read_csv(my_file)),
+  tar_target(my_model, model_function(my_data))
+  tar_quarto(report, "file.qmd", extra_files = "references.bib")     # <2>
+  )
+```
+
+1. Load `tarchetypes` package for quarto support.
+2. Add a quarto target.
+
+## Exercise
+
+* Add a targets workflow to your quarto document.
+* Create a visualization of the pipeline network using `tar_visnetwork()`.
+
+## Assignment 4

week9/targets_example/_targets.R

@@ -0,0 +1,9 @@
+library(targets)
+source("functions.R")
+tar_option_set(packages = c("readr", "dplyr", "ggplot2"))
+list(
+  tar_target(file, "data.csv", format = "file"),
+  tar_target(data, get_data(file)),
+  tar_target(model, fit_model(data)),
+  tar_target(plot, plot_model(model, data))
+)