tudelft_20nov2024.qmd

---
title: "Tips for reproducibility in R"
author: "Sam Langton -- Amsterdam UMC"
subtitle: "<br> <br> <br> TU Delft, 20 November 2024"
format:
  revealjs:
    auto-stretch: false
editor: visual
editor_options: 
  chunk_output_type: console
date-format: long
---

## Who am I?

::: columns
::: {.column width="60%"}
-   Research software consultant at AUMC
-   Research background: spatial patterning of crime
-   R enthusiast
-   Promoting & training open software
:::

::: {.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>
:::
:::

## What is reproducibility?

> "We define reproducible research as work that can be independently recreated from the same data and the same code that the original team used."

<center>

![Source: [The Turing Way](https://book.the-turing-way.org/reproducible-research/overview/overview-definitions)](https://book.the-turing-way.org/_images/reproducible-matrix.jpg){width="500"}

</center>

## A personal tale

<br>

Publishing an [open science replication paper](https://www.tandfonline.com/doi/full/10.1080/15614263.2022.2102494).

<br>

<center>![](img/paper_header.png){width="900"}</center>

## 

<center>![](img/spongebob.png){width="450"}</center>

## A personal tale

<br>

...and I could not reproduce the findings.

<br>

<br>

<br>

(It was fixed, but with [some effort](https://github.com/langtonhugh/demand_detroit) and a bit of luck).

## 

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

## Focus your energy

1.  Script *everything* ([De Graaff., 2021](https://www.dropbox.com/s/vn7rn9m5ht2miol/reproducibility.pdf?e=1&dl=0))
2.  Project folders
3.  Options in RStudio
4.  `renv`
5.  Quarto

## Script *everything*

<br>

"R is a statistics software" `r emo::ji("angry")` Well, partly... but it can (probably) be used for your [entire computational workflow](https://wec.wur.nl/dse/images/11/tidyverse%20workflow.png).

<br>

If you **can** script it, then **script it**.

<br>

RStudio can also support other languages (e.g., [Python](https://posit.co/blog/three-ways-to-program-in-python-with-rstudio/))

## 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.  `renv::init()`
3.  Do some stuff
4.  `renv::snapshot()`
5.  Share with someone
6.  `renv::restore()`
7.  Do some other stuff
8.  Go to point 4...

## A low-effort alternative

<br>

Saving your **session info** is the **least** we can do! 

<br>

<br>

```{r, eval = F, echo = T}
writeLines(capture.output(sessionInfo()), "sessionInfo.txt")`
```

<!-- ## A (better?) alternative: `rix` -->

<!-- <br> -->

<!-- New (to me) but claims to offer advantages over `renv`. -->

<!-- <br> -->

<!-- Have [a read](https://cran.r-project.org/web/packages/rix/index.html) or [watch a tutorial](https://www.youtube.com/watch?v=tM4JrCWZpwA) to find out more. -->

## 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://quarto.org/docs/gallery/).

## Summary

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

-   Script **everything**\
-   Use project files
-   Use `renv` to manage package versions
-   Use Quarto
:::

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

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

## Extras

### Fundamentals:

- Documentation (e.g., README, comments)
- Linear scripts
- Function conflicts

### Bonus points:

- `here` package
- Version control (e.g., GitHub)

## Advertisements!

<center>

<br>

::: columns
::: {.column width="50%"}
![](img/codecheck_logo.svg){width="400" fig-align="center"}

Upcoming event [in Rotterdam](https://codecheck.org.uk/nl/workshop3.html).
:::

::: {.column width="50%"}
![](img/NLRN.png){width="400" fig-align="center"}

Dutch Reproducibility Network event [in Groningen](https://reproducibilitynetwork.nl/event/nlrnsymposium2024/). 
:::
:::

</center>


## Thanks for listening!

<br>

<center>

#### Questions? Suggestions? Comments?

<br>

#### Slides

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

<br>

#### Contact

s.h.langton\@amsterdamumc.nl

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