Kiplinger's Best Value Rankings (Refactored)

kiplingers_refactored.R

# REFACTORED VERSION OF CODE IN GIST: git://gist.github.com/1362541.git

# FUNCTION TO GET TABLE BASED ON COLLEGE TYPE
get_table <- function(college_type){
  require(XML)
  url_base <- "http://www.kiplinger.com/tools/privatecolleges/index.php?table"
  url_1 <- paste(url_base, college_type, sep = "=")
  tab   <- readHTMLTable(url_1, header = TRUE, stringsAsFactors = FALSE)
  return(tab[[1]])
}

# READ TABLES INTO DATA FRAME AND SET NAMES
college_types  <- list(Private = "prv_univ", `Liberal Arts` = "lib_arts")
mydf <- plyr::ldply(college_types, get_table)
mydf <- setNames(mydf, c('type', "rank", "name", "state", "admit_rate", 
  "stud_per_faculty", "4y_grad_rate", "coa", "avg_need_aid", "avg_noneed_aid", 
  "pct_noneed_aid", "avg_grad_debt"))

# CONVERT COLUMNS WITH % AND $ INTO NUMBERS
mydf_cleaned <- sapply(mydf, gsub, pattern = "[%$,]", replacement = "")
mydf_cleaned <- adply(mydf_cleaned, 2, type.convert)