Overview

• Before You Start Using R
• Once You Start Writing R

All references and additional information can be found in this repository

https://github.com/EmilHvitfeldt/oRganized-talk

Use Rstudio

https://www.rstudio.com/products/rstudio/download/#download

Change Settings

Change Appearance

Pane layouts

Pane layouts - My Setup

I like having both source and console open

Pane layouts - My Setup 2

... while still allowing me to have plots/help/viewer open

RStudio Projects

Keep all files from one project together. Use RStudio projects.

• Self contained
• Avoid overlapping
• Project orientated

RStudio Projects - Creation 1 / 4

Keep all files from one project together. Use RStudio projects.

RStudio Projects - Creation 2 / 4

RStudio Projects - Creation 3 / 4

RStudio Projects - Creation 4 / 4

Folder Structure

name_of_project
|--raw_data
    |--WhateverData.xlsx
    |--report_2017.csv
|--output_data
    |--summary2017.csv
|--rmd
    |--01-analysis.Rmd
|--docs
    |--01-analysis.html
    |--01-analysis.pdf
|--scripts
    |--exploratory_analysis.R
    |--pdf_scraper.R
|--figures
    |--weather_2017.png
|--name_of_project.Rproj
|--run_all.R

Folder Structure

Everything has a spot where it belongs.

• Raw data separate from cleaned data
• Reports and scrips are separated
• Generated and imported figures has its own place
• Numbered using 2 digits
• Reusable and easily understandable

library(fs)
folder_names <- c("raw_data", "output_data", "rmd", "docs", 
                  "scripts", "figures")
dir_create(fldr_names)

Paths

library(tidyverse)
# data import
data <- read_csv("/Users/Emil/Research/Health/amazing_data.csv")

Paths

library(tidyverse)
# data import
data <- read_csv("/Users/Emil/Research/Health/amazing_data.csv")

## Error: '/Users/Emil/Research/Health/amazing_data.csv' does not exist.

Only use relative paths, never absolute paths

Introducing the here package.

library(here)
here()

## [1] "/Users/Emil/Research/Health"

library(here)
data <- read_csv(here("amazing_data.csv"))

Naming Things

Naming Things - Files

NO

report.pdf
reportv2.pdf
reportthisisthelastone.pages
Figure 2.png 
3465-234szx.r
foo.R

YES

2018-10-01_01_report-for-cdc.pdf
01_data.rmd
01_data.pdf
02_data-filtering.rmd
02_data-filtering.pdf

Naming Things - Files

We want file names to be "machine readable" and "human readable".

• Avoid spaces, punctuation, special characters and case sensitivity
• Deliberate use of delimiters
• File name should describe the contents of the file
• Put something numeric first
• Left pad numbers with zeroes
• Use ISO 8601 standard for dates (YYYY-MM-DD)

Naming Things - Files

library(fs)
dir_ls("data/", regexp = "health-study")

## 2018-02-23_health-study_power-100_group-A1.csv
## 2018-02-23_health-study_power-100_group-B1.csv
## 2018-02-23_health-study_power-100_group-C1.csv
## 2018-02-23_health-study_power-200_group-A1.csv
## 2018-02-23_health-study_power-200_group-B1.csv
## 2018-02-23_health-study_power-200_group-C1.csv

stringr::str_split_fixed(x, "[_\\.]", 5)

##      [,1]         [,2]           [,3]        [,4]       [,5] 
## [1,] "2018-02-23" "health-study" "power-100" "group-A1" "csv"
## [2,] "2018-02-23" "health-study" "power-100" "group-B1" "csv"
## [3,] "2018-02-23" "health-study" "power-100" "group-C1" "csv"
## [4,] "2018-02-23" "health-study" "power-200" "group-A1" "csv"
## [5,] "2018-02-23" "health-study" "power-200" "group-B1" "csv"
## [6,] "2018-02-23" "health-study" "power-200" "group-C1" "csv"

Naming Things - Files

library(fs)
dir_ls("data/", regexp = "health-study")

## 2018-02-23_health-study_power-100_group-A1.csv
## 2018-02-23_health-study_power-100_group-B1.csv
## 2018-02-23_health-study_power-100_group-C1.csv
## 2018-02-23_health-study_power-200_group-A1.csv
## 2018-02-23_health-study_power-200_group-B1.csv
## 2018-02-23_health-study_power-200_group-C1.csv

library(tidyverse)
map_df(dir_ls("data/", regexp = "health-study"), read_csv)
# or
dir_ls("data/", regexp = "health-study") %>%
  map_df(read_csv)

Naming Things - Objects

• Only use lowercase letters, numbers, and _
• Use names that are not jargony, weight instead of K
• Use informative names

# Bad
df
e
tuningVar
# Good
health_data
error
tuning_var

What To Avoid - attach()

Never use attach()

attach(mtcars)
mean(mpg)

## [1] 20.09062

Loading lots of names into the search path, ambiguous selections.

What To Avoid - attach()

Never use rm(list=ls())

Will not reset packages, reset options, ...

Restart the R session

CTRL+SHIFT+F10 for Windows
CMD+SHIFT+ALT+F10 for Mac OS

R Markdown

Most of your documents can be written in R Markdown.

Using the basics of markdown with the addition of R code chunks.

R Markdown

R Markdown documents versus R scripts

Use R scripts for simple self contained tasks.

source() R scripts into your R Markdown document where you will do analyses, visualizations and reporting.

R Markdown

- 01-import.R
- 02-clean-names.R
- 03-tidy.R
- etc

Include at the start of R Markdown file

{r load_scripts, include = FALSE}
library(here)
source(here("scripts", "01-import.R"))
source(here("scripts", "02-clean-names.R"))
source(here("scripts", "03-tidy.R"))

Naming Chunks

Names can be placed after the comma

```{r, chunk-label, results='hide', fig.height=4}

or before

```{r chunk-label, results='hide', fig.height=4}

In general it is recommended to use alphabetic characters with words separated by - and avoid other characters. - Yihui Xie

Naming Chunks

• Make navigating the R Markdown document easier
• Make your R Markdown easier to understand
• Clarifies error reports or progress of knitting
• Caching when moving chunks around
• Feels good

Setup Chunk

In a fresh R Markdown document you see this

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)

The setup chuck is run before another code - use to your advantage

Chunk Options

Defaults will rarely work for you 100% of the time.

Set echo = TRUE, knit document with code.
Set echo = FALSE, reknit document with the code hidden.

Chunk Options

```{r setup, include=FALSE}
knitr::opts_chunk$set(fig.path = "figures/")

Will save all figures in the the folder figures with the chunk name.

Styling Code

Use consistent style when writing code

http://style.tidyverse.org/

All about preferences but keep it consistent!!!

Keep .Rprofile Clean

Your project contains a file called .Rprofile.

This file runs first in every session. Think of it as configuration file.

options(stringsAsFactors = FALSE)
options(max.print = 100)

PANIC!!! Code will only work for you!
Avoid putting output altering code in .Rprofile!

Comment Your Code

• Functions: Arguments and purpose
• Code: What or why, NOT how

# Takes a data.frame (data) and replaces the columns with the names
# (names) and converts them from factor variable to character 
# variables. Keeps characters variables unchanged.
factor_to_text <- function(data, names) {
  for (i in seq_along(names)) {
    if(is.factor(data[, names[i], drop = TRUE]))
      data[, names[i]] <- as.character.factor(data[, names[i], 
                                                   drop = TRUE])
  }
  data
}

Should You Write a Function?

Once you’ve written the same code 3 times, write a function

mtcars %>% 
  mutate(mpg = mpg / 2.5 + 2,
         cyl = cyl / 2.5 + 2,
         hp = hp / 2.5 + 2)

You should have added 3 instead of 2

mtcars %>% 
  mutate(mpg = mpg / 2.5 + 2,
         cyl = cyl / 2.5 + 3,
         hp = hp / 2.5 + 3)

Should You Write a Function?

Once you’ve written the same code 3 times, write a function

scale_fun <- function(x) x / 2.5 + 3
mtcars %>% 
  mutate(mpg = scale_fun(mpg),
         cyl = scale_fun(cyl),
         hp = scale_fun(hp))

Testing

Use tests in your pipeline to check assumptions.

library(tidyverse)
data <- fancy_gov_data_api("health", year = 2018, month = 10)
data %>% 
  group_by(county) %>% 
  summarize(happiness = (- death + love) / population)

Testing

testhat is primarily used for unit testing in R packages.

Contains large collection of expect_ functions.

• expect_equal
• expect_equivalent
• expect_error
• expect_length
• expect_named
• ...

Testing

Use tests in your pipeline to check assumptions.

library(tidyverse)
library(testthat)
data <- fancy_gov_data_api("health", year = 2018, month = 10)
expect_length(unique(data$county), 3007)
expect_gt(mean(data$love == NA), 0.1)
data %>% 
  group_by(county) %>% 
  summarize(happiness = (- death + love) / population)