Best Practices in Teaching R

Erin Grand
November 28, 2017

Previously...

  • System that worked, but was hard and long
  • Excel workbooks to process data
  • Long detailed process to download, clean and process data

Why Choose R?

  • Free: R and Rstudio are free and open sourced!
  • Reproducible: Easy to run scripts again with updated data.
  • Collaboration: Results are easy to share as a PDF.
  • Manipulation: Data manipulation is much easier due to the tidyverse.
  • Understandable: You can easily follow along in someone else's code.
  • Spot Checking: You can QC yourself along the way, instead of all at the end.
  • Help: Finding answers to your programming questions is a quick google away.

Pain Point: Time!

It takes serious TIME to move everyone and everything to code.

  • Dedicate time to designing and giving PD
  • Steep learning curve outside of PD
    • Required 1:1 training and code reviews
  • Redesign of project plans for analyses to use automation
    @astroeringrand

Training Practices

  • Materials such as Slides + Cheat sheets available online
  • Following examples in R for Data Science
  • Practice questions and examples
  • Pair coding
  • Code reviews
  • Use of my own mistakes as examples of what not to do

R Scope and Sequence

  • Base R: Data structures, data types, basic of coding, simple stats
  • Tidyverse: Tibbles, dplyr, %>%
  • Data Cleaning: janitor, tidyr
  • Data Checks: assertr
  • Functions
  • PURRR
  • More advanced statistics (i.e regressions)

Example: Data Types

Data type Example
Integer 1
Logical TRUE
Numeric 1.1
String / character "Red"
Factor (enumerated string) "Amber" or 2 in c("Red","Amber","Green")
Complex i
Date 2015-04-24
NA NA

Example: Mutate

Besides selecting sets of existing columns, it's often useful to add new columns that are functions of existing columns. That's the job of mutate().

library(tidyverse)
library(fivethirtyeight)

avengers <- mutate(fivethirtyeight::avengers, 
       death1 = if_else(!is.na(death1) & death1, 1, 0 ),
       death2 = if_else(!is.na(death2) & death2, 1, 0 ),
       death3 = if_else(!is.na(death3) & death3, 1, 0 ),
       death4 = if_else(!is.na(death4) & death4, 1, 0 ),
       death5 = if_else(!is.na(death5) & death5, 1, 0 ),
       total_deaths = death1 + death2 + death3 + death4 + death5)

Task: Create a column called Pass that describes if the student is predicted to pass the exam (1) or not (0).

Cheat Sheets