gendercause.R

# This function was written for working with this dataset:
# https://spark-public.s3.amazonaws.com/compdata/data/Baltimore_homicides.zip
# It extracts all cases of a given cause and returns a data frame with two
# coulmns Cause and Gender.

gendercause <- function(data, cause) {
  data <- readLines("homicides.txt")
  if (cause == "asphyxiation") {
    asph <- grep("[Cc]ause: [Aa]sphyxiation", data)
    female <- grep("[Ff]emale", data[asph])
    male <- grep(" [Mm]ale", data[asph])
    cause <- character(0)
    for (i in asph) {
      i <- "asphyxiation"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
  else if (cause == "other") {
    oth <- grep("[Cc]ause: [Oo]ther", data)
    female <- grep("[Ff]emale", data[oth])
    male <- grep(" [Mm]ale", data[oth])
    cause <- character(0)
    for (i in oth) {
      i <- "other"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
  else if (cause == "blunt force") {
    bf <- grep("[Cc]ause: [Bb]lunt.[Ff]orce", data)
    female <- grep("[Ff]emale", data[bf])
    male <- grep(" [Mm]ale", data[bf])
    cause <- character(0)
    for (i in bf) {
      i <- "blunt force"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
  else if (cause == "shooting") {
    shoot <- grep("[Cc]ause: [Ss]hooting", data)
    female <- grep("[Ff]emale", data[shoot])
    male <- grep("[^t] [Mm]ale", data[shoot])
    cause <- character(0)
    for (i in shoot) {
      i <- "shooting"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
  else if (cause == "stabbing") {
    stab <- grep("[Cc]ause: [Ss]tabbing", data)
    female <- grep("[Ff]emale", data[stab])
    male <- grep(" [Mm]ale", data[stab])
    cause <- character(0)
    for (i in stab) {
      i <- "stabbing"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
  else if (cause == "unknown") {
    unknown <- grep("[Cc]ause: [Uu]nknown", data)
    female <- grep("[Ff]emale", data[unknown])
    male <- grep(" [Mm]ale", data[unknown])
    cause <- character(0)
    for (i in unknown) {
      i <- "unknown"
      cause <- c(i, cause)
    }
    genfem <- character(0)
    for (i in female) {
      i <- "female"
      genfem <- c(i, genfem)
    }
    genmal <- character(0)
    for (i in male) {
      i <- "male"
      genmal <- c(i, genmal)
    }
    gentotal <- c(genfem, genmal)
    output <- data.frame(Cause = cause, Gender = gentotal)
    return(output)
  }
}