Data processing for surveys with multiple choice answers put in a column with seperated string-values like "R;Python;C"

expand multiple choice to bool columns.R

# Suvery tools often put multiple choice questions’s answers as a single value with concatenated strings: 
# If asking which programming languages from a list people use, tools will output "C;Python;R" if users selected these options.
# These will be put in a column representing the question’s answers. However, this format is hard to work with.
#
# The code below creates a %>%-able function "stringColToBoolMatches"
#
# parameters: df <tibble>, stringColName<character>, name of the column with the answers, separator<character> e.g. ";"
# returning: a new dataframe that keeps the original column but expands its values to boolean columns, one for each multiple choice option
#
# the columns are named after the unique answer option and can be NA (original value was NA), True (value present) or False (value not present)
# e.g. if the column "lang" had a value like "C;Python;R", in the same row the columns "lang_C", "lang_Python", "lang_R" will be True, 
# but "lang_Java", "lang_go", etc. will be False. 

library(tidyverse)

# parameters: df <tibble>, columnName<character>, separator<character>
# returns: <character> vector 
getUniqueValuesFromStringColumn <- function(df,columnName,separator){
  column <- df[[columnName]]
  paste(column, collapse = separator) %>%
  strsplit(split=separator) %>%
  unlist %>%
  unique%>%
  sort(decreasing = TRUE)
}

# parameters: answersValue<character> e.g. "C; python; Java", optionToCheck<character> e.g. "Java", separator<character> e.g. ";"
# returns: <boolean>
isInAnswerOptions<- function(answersValue,optionToCheck,separator){
  if(is.na(answersValue)) return(NA)
  answerList <- strsplit(answersValue,split=separator)
  optionToCheck %in% unlist(answerList) #unlist cause %in% needs vector
}


# parameters: df <tibble>, stringColName<character>, name of the column with the answers, answerOption<character> e.g. "Java", separator<character> e.g. ";"
# returns: <tibble>
addBoolColForAnswerOption <- function(df,stringColName,answerOption,separator){
  stringCol <- df[[stringColName]]
  inAnswerOptionsPrefilled <- partial(isInAnswerOptions, optionToCheck=answerOption, separator=separator)
  newdf <- mutate(df,"{stringColName}_{answerOption}":=map_lgl(stringCol,inAnswerOptionsPrefilled), .after=ends_with(stringColName))
} 

# df <tibble>, stringColName<character>, name of the column with the answers, separator<character> e.g. ";"
# returns: <tibble>
stringColToBoolMatches <- function(df,stringColName,separator){
  uniqueValues <- getUniqueValuesFromStringColumn(df,stringColName,separator)
  for (subColValue in uniqueValues){
    df <- addBoolColForAnswerOption(df,stringColName,subColValue,separator)
  }
  df
}

# can be used like so: 
survey_results_public <- read_csv("C:/Users/jan/Downloads/survey_results_public.csv") # https://insights.stackoverflow.com/survey

so_forAnalysis <- survey_results_public %>%
stringColToBoolMatches(stringColName = "LearnCode",separator = ";")%>%
stringColToBoolMatches(stringColName = "LearnCodeOnline", separator = ";")%>%
stringColToBoolMatches(stringColName = "LanguageHaveWorkedWith", separator = ";")%>%
stringColToBoolMatches(stringColName = "CodingActivities", separator = ";")%>%
select(-ends_with("_NA")) # NAs are currently put in their own column, I guess cause they are cast to strings. Might fix at some point.