Troubleshooting

class: center, middle, inverse, title-slide

# Troubleshooting
## and problem solving in R
### Alec L. Robitaille
### Wildlife Evolutionary Ecology Lab Memorial University
### 2022-04-18

---

.left-column[
 
### **Discussion**
### Prevention
### Simplifying
### and other strategies
### Resources
]

---

.left-column[
 
### **Discussion**
### Prevention
### Simplifying
### and other strategies
### Resources
]

#### What are your strategies for

...troubleshooting errors?

...problem solving?

...learning a new package?

...starting a new analysis?

---

.left-column[
 
### Discussion
### **Prevention**
### Simplifying
### and other strategies
### Resources
]

#### First, some preventative measures

Adjust RStudio settings [[1]](https://csgillespie.github.io/efficientR/set-up.html#rstudio-options)

Use an RStudio project, relative file paths

Keep R, R packages, and OS updated

---

.left-column[
 
### Discussion
### **Prevention**
### Simplifying
### and other strategies
### Resources
]

#### Extras

Track package versions with `renv`

Reproducible workflows with `targets`

Use Git

???
renv tracks package versions for each project, so global updates to packages 
won't break eg. older projects

Git helps with the feedback cycle of coding -> tracked changes -> commit -> coding
Also, Git lets you create branches to explore alternatives, without 
breaking your current working versions

A workflow defined in targets ensures steps are only rerun when they are outdated 
and helps branching by eg. individuals or study sites

---

.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

####  Overview

Simplify the complexity of the task

...to identify source of the error

...to clarify steps of the analysis

...to learn new functions or packages

???

Tasks: troubleshooting, learning a new package, starting a new analysis...

We'll first look at making minimal data, then using that data to
learn a new package, then finally taking a complex script + data and
reducing it to a minimum example

---

.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Data

Goal: minimal and representative data

Option 1 - available example data

Option 2 - build your own

???

Representative: retains variables, types and grouping levels of original data

---

.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Available data

.panelset[
.panel[.panel-name[Penguins]

```r
library(palmerpenguins)
data(penguins)
nrow(penguins)
```

```
## [1] 344
```

```r
colnames(penguins)
```

```
## [1] "species"           "island"            "bill_length_mm"   
## [4] "bill_depth_mm"     "flipper_length_mm" "body_mass_g"      
## [7] "sex"               "year"
```
]
.panel[.panel-name[Cars]

```r
data(mtcars)
nrow(mtcars)
```

```
## [1] 32
```

```r
colnames(mtcars)
```

```
##  [1] "mpg"  "cyl"  "disp" "hp"   "drat" "wt"   "qsec" "vs"   "am"   "gear"
## [11] "carb"
```
]
.panel[.panel-name[Etc]

```r
# Run data() to see a list of available data
data()

# for spatial data, see the spData package
data(package = 'spData')
```
]
]
---

.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Exercise: build your own data

1) Take your real data

2) List  
-columns  
-data types  
-units  
-limits  
-factors or levels

3) Generate a representative and minimal fake data

.footnote[
[`R/simplify-data.R`](https://github.com/robitalec/workshops/tree/master/2022-04-18-troubleshooting/R/simplify-data.R)
]

---
.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Learning a new package

Use a simple dataset

to focus on learning functions

without complexity of real data or specific analyses

---
.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Exercise: learn a new package

1) Use a minimal dataset

2) Open package website or documentation

3) Fill the [template](https://github.com/robitalec/workshops/tree/master/2022-04-18-troubleshooting/R/learn-package-template.R) (see the [example](https://github.com/robitalec/workshops/tree/master/2022-04-18-troubleshooting/R/learn-package-chk.R))

4) Learn a few functions!

.footnote[
Pick any package or one from this list:  
-[`ggplot2`](https://ggplot2.tidyverse.org/)  
-[`sf`](https://r-spatial.github.io/sf/)  
-[`chk`](https://poissonconsulting.github.io/chk/)  
-[`anytime`](https://eddelbuettel.github.io/anytime/)    
-[`patchwork`](https://patchwork.data-imaginist.com/)  
]

---
.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Troubleshooting

To troubleshoot an error,  
simplify the data and analyses

Isolate the erroring function and   
use a minimal dataset

If the error is not reproduced, check differences 
in data and/or use of functions

If error is reproduced, check GitHub Issues, SO, try on another
computer, or contact package authors.

???
A MWE that reproduces an error is ~~ proof of the error

---
.left-column[
 
### Discussion
### Prevention
### **Simplifying**
### and other strategies
### Resources
]

#### Exercise: troubleshooting

Given a complex script (e.g.  [1](https://github.com/wildlifeevoeco/SocCaribou/blob/master/scripts/2-HomeRangeAnalysis.R), [2](https://github.com/robitalec/ScaleInMultilayerNetworks/blob/master/scripts/03-temporal-layers.R), [3](https://github.com/wildlifeevoeco/study-area-figs/blob/master/R/10-riding-mountain-prep.R), [4](https://github.com/robitalec/spatsoc-application-paper/blob/master/scripts/figure-2-group-pts.R)),  
build a reprex

1) Start a new script

2) Only load essential packages

3) Use a minimal dataset

4) Isolate the erroring function

5) Reproduce the error  
or find your mistake

.footnote[
[`reprex` vignette](https://reprex.tidyverse.org/articles/reprex-dos-and-donts.html)  
[SO reprex how to](https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example/16532098)
]

???
Ideally, bring your own script

Say you hit an error halfway through some complex analysis...

`reprex_selection`

---

.left-column[
 
### Discussion
### Prevention
### Simplifying
### **and other strategies**
### Resources
]

Explore the traceback [[1]](https://adv-r.hadley.nz/debugging.html#traceback)

Read (and reread) documentation

Restart R, clear the environment

Update R, R packages, OS

Check your .Rprofile, .Renviron
[[2]](https://csgillespie.github.io/efficientR/set-up.html#an-overview-of-rs-startup-files=), [[3]](https://usethis.r-lib.org/reference/edit.html)

GitHub Issues [(e.g. [4])](https://github.com/rstudio/bookdown/issues/1263)  
Stack Overflow [(e.g. [5])](https://stackoverflow.com/questions/42149427/group-membership-by-pairwise-criteria-matching)

Ask for help!

---

.left-column[
 
### Discussion
### Prevention
### Simplifying
### and other strategies
### **Resources**
]

-[`reprex`: introduction](https://reprex.tidyverse.org/articles/reprex-dos-and-donts.html)  
-[SO: Making a great R reproducible example](https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example/16532098)  
-[`reprex`: datapasta](https://reprex.tidyverse.org/articles/datapasta-reprex.html)

-[Advanced R: Debugging](https://adv-r.hadley.nz/debugging.html0)

-[Description of startup files](https://csgillespie.github.io/efficientR/set-up.html#an-overview-of-rs-startup-files=)  
-[Editing startup files](https://usethis.r-lib.org/reference/edit.html)

-[`targets` manual](https://books.ropensci.org/targets/)  
-[`renv` documentation](https://rstudio.github.io/renv/)

-[Efficient R: Version control](https://csgillespie.github.io/efficientR/collaboration.html#version-control)  
-[Happy Git with R](https://happygitwithr.com/)