data-raw/0-clean.R

In hadley/modelr: Modelling Functions that Work with the Pipe

# 0-clean.R
# setwd() to this folder

# Data was downloaded from https://www.nlsinfo.org/investigator/. Variables were
# selected manually with the web api. You can recreate this data set by
# selecting the variables with the following reference numbers: A0002500,
# R0000100, R0214700, R0214800, R0618301, R1774100, R1774200, T3212900,
# T3955000, T3955100, T3955200, T3977400, T3979400, T4112900, T4113200 from the
# NLSY79 (1979-2012) study

# The clean script below is adapted from the clean script provided by NLS
# investigator, NLSY79-2012-bookvars.R

new_data <- read.table("NLSY79-2012-bookvars.dat", sep = " ")
varlabels <- c(
  "case", "id", "race", "sex", "afqt", "hair", "eyes", "education",
  "weight", "feet", "inches", "income", "earnings", "marital", "age"
)
names(new_data) <- varlabels

# Handle missing values
new_data[new_data == -1] <- NA # Refused
new_data[new_data == -2] <- NA # Dont know
new_data[new_data == -3] <- NA # Invalid missing
new_data[new_data == -4] <- NA # Valid missing
new_data[new_data == -5] <- NA # Non-interview

# If there are values not categorized they will be represented as NA
vallabels <- function(data) {
  data$race <- factor(data$race, levels = c(1.0, 2.0, 3.0), labels = c("hispanic", "black", "other"))
  data$sex <- factor(data$sex, levels = c(1.0, 2.0), labels = c("male", "female"))
  data$hair <- factor(data$hair, levels = c(1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0), labels = c("light_blond", "blond", "light_brown", "brown", "black", "red", "grey"))
  data$eyes <- factor(data$eyes, levels = c(1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0), labels = c("light_blue", "blue", "light_brown", "brown", "black", "green", "hazel", "grey", "other"))
  data$education[data$education == 95] <- NA
  data$marital <- factor(new_data$marital, levels = c(0.0, 1.0, 2.0, 3.0, 6.0), labels = c("single", "married", "separated", "divorced", "widowed"))
  return(data)
}

categories <- vallabels(new_data)

# standardize heights
categories$feet[categories$inches > 12 & !is.na(categories$inches)] <- 0

library(dplyr)

heights <-
  categories %>%
  mutate(height = inches + feet * 12, afqt = afqt / 1000) %>%
  filter(!is.na(income), height < 90, height > 40) %>%
  select(id, income, height, weight, age, hair, eyes, marital, sex, education, afqt)

saveRDS(heights, file = "heights.RDS")