data-raw/onet_activities_read.R

In djmorris1989/onetmappinguk: Map US O-NET to the UK SOC-2010

#' Read in the work activites data
#'
#' This function loads in the text files from the O*NET database releases, with one selected for each year from
#' 2004 to 2019. It cleans the data to produce a consistantly labelled panel - though the occuptions change over
#' time due to multiple standard occupational classification changes
#'
#'
#' @return `O*NET-SOC Code` - the occupational code
#' @return `Element Name` - the name of the activity
#' @return `source` - who provided the activity rating
#' @return `year` - year in which the underlying database was released
#' @return `Incumbent` - dummy variable equal to 1 if the source is a job incumbent
#' @return `Data Value IM` - the average value of the 1-5 activity importance rating across respondents
#' @return `Standard Error IM` - standard error of the importance value
#' @return `Data Value LV` - the average value of the 1-5 activity level rating across respondents
#' @return `Standard Error LV` - standard error of the level value
#' @export
onet_activities_read <- function() {

  data.2004 <- data.table::fread("data/onet databases/db_60/db_60/WorkActivity.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2005 <- data.table::fread("data/onet databases/db_80/db_80/WorkActivity.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2006 <- data.table::fread("data/onet databases/db_10_0/db_10_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2007 <- data.table::fread("data/onet databases/db_12_0/db_12_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2008 <- data.table::fread("data/onet databases/db_13_0/db_13_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2009 <- data.table::fread("data/onet databases/db_14_0/db_14_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2010 <- data.table::fread("data/onet databases/db_15_0/db_15_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2011 <- data.table::fread("data/onet databases/db_16_0/db_16_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2012 <- data.table::fread("data/onet databases/db_17_0/db_17_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2013 <- data.table::fread("data/onet databases/db_18_0/db_18_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2014 <- data.table::fread("data/onet databases/db_19_0/db_19_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2015 <- data.table::fread("data/onet databases/db_20_0/db_20_0/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2016 <- data.table::fread("data/onet databases/db_21_0_text/db_21_0_text/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2017 <- data.table::fread("data/onet databases/db_22_0_text/db_22_0_text/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2018 <- data.table::fread("data/onet databases/db_23_0_text/db_23_0_text/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]
  data.2019 <- data.table::fread("data/onet databases/db_24_0_text/db_24_0_text/Work Activities.txt")[,c("O*NET-SOC Code","Element Name","Scale ID","Data Value","Standard Error","Domain Source")]

  data.2004$year = 2004
  data.2005$year = 2005
  data.2006$year = 2006
  data.2007$year = 2007
  data.2008$year = 2008
  data.2009$year = 2009
  data.2010$year = 2010
  data.2011$year = 2011
  data.2012$year = 2012
  data.2013$year = 2013
  data.2014$year = 2014
  data.2015$year = 2015
  data.2016$year = 2016
  data.2017$year = 2017
  data.2018$year = 2018
  data.2019$year = 2019

  ### append into one data frame
  data <- dplyr::bind_rows(data.2004,data.2005,data.2006,data.2007,data.2008,data.2009,data.2010,data.2011,
                           data.2012,data.2013,data.2014,data.2015,data.2016,data.2017,data.2018,data.2019) %>%
    dplyr::rename("source"="Domain Source") %>%
    dplyr::mutate(Incumbent = ifelse(source=="Incumbent",1,0))


  ### reshape the data so there are separate variables for levels and importance

  rawdata_activities <- reshape(data=data,
                      idvar=c("O*NET-SOC Code","Element Name","year"),
                      timevar="Scale ID",
                      v.names=c("Data Value","Standard Error"),
                      direction="wide",
                      sep=" ") %>%
    dplyr::rename("element"="Element Name",
                  "data_I"="Data Value IM",
                  "data_L"="Data Value LV",
                  "se_I" = "Standard Error IM",
                  "se_L" = "Standard Error LV")

  return(reshaped)

}