reprohack_22may2024.qmd

---
title: "Reproducible workflows in R"
author: "Sam Langton -- Amsterdam UMC"
date: "22 May, 2024"
format:
  revealjs:
    auto-stretch: false
editor: visual
editor_options: 
  chunk_output_type: console
date-format: long
---

## Who am I?

::: columns
::: {.column width="60%"}
-   Software/data consultant in RDM
-   Research background: spatial patterning of crime
-   R enthusiast
-   Here to make reproducibility easier!
:::

::: {.column width="40%"}
<center>

![](https://github.com/langtonhugh/website_academic/blob/master/content/authors/admin/avatar.jpg?raw=true){width="200"}

![](img/crime_map.png){width="300"}

</center>
:::
:::

## Reproducibility: why bother?

-   Yourself
-   Your collaborators
-   The scientific community
-   Stakeholders (e.g., the public)

## Tell me!

<br> <br>

<center>

Why does your (or your colleagues') code break?

## Tell me!

<br> <br>

<center>

Have you ever ended up with different results to the last run?

<Br>

Did you know why?

<!-- ## Tell me! -->

<!-- <br> <br> -->

<!-- <center>How do you ~~ensure~~ maximize the chances that your analyses are reproducible?</center> -->

## 

<center>![](img/drake.png){width="800"}</center>

## Focus your energy

1.  Project folders
2.  Options in RStudio
3.  The `here` package
4.  `renv`
5.  Quarto

## Project folders

<br> <br>

<center>

![Source: [Jenny Bryan](https://www.tidyverse.org/blog/2017/12/workflow-vs-script/)](img/setwd.png){width="700"}

</center>

## Project folders

<br>

<center>

![](img/project_screenshot.png){width="700"}

<br>

![](img/folder_screenshot.png){width="800"}

</center>

## Project folders

Avoids this:

```{r, eval = FALSE, echo = TRUE}
setwd("C:/SamL/my_projects/ridiculous_folder_name/data")

raw_data_df <- read_csv("raw_records.csv")
```

<br> And this:

```{r, eval = FALSE, echo = TRUE}
raw_data_df <- read_csv("C:/SamL/my_projects/ridiculous_folder_name/data/raw_records.csv")
```

<br>

But instead:

```{r, eval = FALSE, echo = TRUE}
raw_data_df <- read_csv("data/raw_records.csv")
```

</center>

<!-- ## Options in RStudio -->

<!-- <br> -->

<!-- <br> -->

<!-- <center> -->

<!-- `Ctrl + shift + F10` -->

<!-- <br> -->

<!-- `Ctrl + A` -->

<!-- <br> -->

<!-- `Ctrl + Enter` -->

<!-- </center> -->

## Options in RStudio

<br>

<center>

![](img/global_options.png){width="700"}

*Workspaces are pure evil. Turn them off.*

## `here` package

Bad:

```{r, eval = FALSE, echo = TRUE}
raw_data_df <- read_csv("C:/SamL/my_projects/ridiculous_folder_name/data/raw_records.csv")
```

<br>

Good:

```{r, eval = FALSE, echo = TRUE}
raw_data_df <- read_csv("data/raw_records.csv")
```

<br>

Even better:

```{r, eval = FALSE, echo = TRUE}
raw_data_df <- read_csv(
  here("data", "raw_records.csv")
)
```

## `renv`

<center>

![](img/renv.png){width="400"}

[Read this](https://rstudio.github.io/renv/).

</center>

## `renv` demonstration

## `renv` demonstration

1.  Create a project file
2.  Do some stuff
3.  `renv::init()`
4.  Do some more stuff
5.  `renv::snapshot()`
6.  Share with someone
7.  `renv::restore()`
8.  Do some other stuff
9.  Go to point 5...

## Quarto

<br>

```{r}
effiency_number <- 200
```

Using Quarto allows you to make cool documents (and slideshows) while automating your results, making you approximately `r effiency_number`% more efficient.

## Quarto

<br>

```{r}
effiency_number <- 200
```

Using Quarto allows you to make cool documents (and slideshows) while automating your results, making you approximately `r effiency_number`% more efficient.

![](img/quarto.png){width="1000"}

## Quarto

<br>

```{r}
effiency_number <- 200
```

Using Quarto allows you to make cool documents (and slideshows) while automating your results, making you approximately `r effiency_number`% more efficient.

![](img/quarto.png){width="1000"}

It also makes everything [look better](https://github.com/danieltomasz/hikmah-academic-quarto?tab=readme-ov-file).

## Summary

::: columns
::: {.column width="50%"}
DOs

-   Use project files
-   Use `here` for working directories
-   Use `renv` to manage package versions
-   Use Quarto
:::

::: {.column width="50%"}
DON'Ts

-   Use `setwd()`
-   Save workspaces
-   Manually copy results
:::
:::

## Thanks for listening!

<br>

<br>

<center>

Questions? Suggestions? Want advice?

<br>

Slides: [https://github.com/langtonhugh/reprod_r]()

<br>

Sam Langton

s.h.langton\@amsterdamumc.nl

[www.samlangton.info](www.samlangton.info)

## Extras

#### Assumed:

-   Script *everything* ([De Graaff., 2021](https://www.dropbox.com/s/vn7rn9m5ht2miol/reproducibility.pdf?e=1&dl=0))
-   Documentation
-   Linear scripts
-   Function conflicts
-   Comment your code

#### Bonus points:

-   Version control (e.g., GitHub)