Tidyverse Practice

Duration: ~45 Minutes

Learning Objectives

Manage a project
Prepare a dataset for analysis
Save a script as a record of your data prep

Download the data

Fatality Analysis Reporting System, NYS, 2015

Workshop dataset

Source: FARS website

Copy the data

Navigate to your project folder
Create 2 folders: data and R
Copy your downloaded CSV to data

Create a new script file

Go to: "File" > "New File" > "R Script"

Create a header

Simple ...


# Author: Abby
# Date: 19 April 2019
# Project: Data management with tidyverse workshop
# Purpose: To practice piping and dplyr
# Data source: FARS person level data, 2015

... or click the down arrow for elaborate ...


#========================================================================#
# Author:  Abigail Stamm (GitHub ID: ajstamm)                            #
# Date:    Friday, 19 April 2019                                         #
# Project: Data management with tidyverse, an R workshop offered by the  #
#          NYS DOH Epidemiology and Biostatistics Community of Practice  #
# Purpose: To practice basic piping and common dplyr commands            #
# Data source: FARS person level data, 2015, https://www.nhtsa.gov/      #
#              research-data/fatality-analysis-reporting-system-fars     #
#========================================================================#

Load packages

Load tidyverse.
It will auto-load dplyr.


library(tidyverse)

Load data

Use read.csv() (base R) or read_csv() (tidyverse).
Differences between them are beyond the scope of this workshop.


fars <- read_csv("raw_data/fars2015nys_person.csv", 
                 col_types="ccccccnccccccccc")

Options vary. To see them, query ?.

Subset data

Select only observations in Albany (where COUNTY = 1).
FARS uses FIPS codes.


my_fars <- fars %>% filter(COUNTY == 1)

Keep only relevant variables

Drop the following variables:

DOA, SEAT_POS, STATE


my_fars <- my_fars %>%
  select(-DOA, -SEAT_POS, -STATE)

Rename your variables

Change the following variable names:

ST_CASE to case_num
PER_NO to person_num
VEH_NO to vehicle_num


my_fars <- my_fars %>% rename(case_num = ST_CASE,
                              person_num = PER_NO,
                              vehicle_num = VEH_NO)

Side Note: `ifelse()`

For conditional statements

only 2 conditions
chain condition pairs


legs <- c(0,2,0,4,2,6,4,8,6,8)
y <- ifelse(legs == 2, "bird", 
            "not bird")

Try it yourself.


z <- ifelse(legs == 0, "snake", 
     ifelse(legs == 2, "bird", 
            "other"))

Create new variables

Create a logical `driver` variable, where TRUE means the person was a driver (PER_TYPE = 1).
Create a categorical `agegroup` variable to identify if the person was:

child (12 and below)
adolescent (13-19)
young adult (20-29)

middle-aged (30-64)
older adult (65 and up)
unknown (999)

Click the down arrow for the code.

Create new variables code


my_fars <- my_fars %>% 
  mutate(
    driver = (PER_TYP == 1),
    agegroup = ifelse(AGE == 999, NA,
               ifelse(AGE < 13, "child",
               ifelse(AGE < 20, "adolescent",
               ifelse(AGE < 30, "young adult",
               ifelse(AGE < 65, "middle-aged",
               ifelse(AGE >= 65, "older adult", NA))))))
    )

Rearrange the data

Arrange the data by RACE and HISPANIC.


my_fars <- my_fars %>% arrange(RACE, HISPANIC)

Check the first few rows. What do you notice?

Put the code together

Combine all steps.

load the data
subset only Albany County
drop state, seat position, index, and dead on arrival
rename vehicle, case, and person number
create driver and age group
sort by race and ethnicity

Click the down arrow for the code.


fars <- read_csv("raw_data/fars2015nys_person.csv", 
                 col_types="ccccccnccccccccc")
my_fars <- fars %>% 
  filter(COUNTY == 1) %>%
  select(-DOA, -SEAT_POS, -STATE) %>%
  rename(case_num = ST_CASE, person_num = PER_NO, 
         vehicle_num = VEH_NO) %>%
  mutate(driver = (PER_TYP == 1),
         agegroup = ifelse(AGE == 999, NA,
                    ifelse(AGE < 13, "child",
                    ifelse(AGE < 20, "adolescent",
                    ifelse(AGE < 30, "young adult",
                    ifelse(AGE < 65, "middle-aged",
                    ifelse(AGE >= 65, "older adult", 
                    NA))))))
  ) %>% arrange(RACE, HISPANIC)

Save your data

Save to data. Use a meaningful name.


write.csv(my_fars, "data/fars2015nys_AlbanyCounty_person.csv", 
          row.names = FALSE)

And Now You Know!

Next up: Summarizing data