corr.R

corr <- function(directory, threshold = 0) {
  setwd(directory)) ## setting the directory
  correlationVector = NULL ## initializing the correlation matrix
  #Looping through ALL the directory's files 
  for (i in 1:332)
  {
    ## Due to the format of the filename, i.e 001, 010  instead of 1, 10. I became aware that the following method works but not efficient, 
    ## but at the time of the completion of this assignment, it was the only way I knew how to do it.           
    if (i <10) { 
      data <- read.csv(
        paste("0","0", as.character(i), ".csv", sep=""),  ## for example, if 'id' =7, we get 007.csv
        header = T, 
        na.strings=c("NA","NaN", " ")
      )
    }
    else if (i>=10 & i<100) { 
      data <- read.csv(
        paste("0", as.character(i), ".csv", sep=""),  ## for example, if 'id' = 17, we get 017.csv
        header = T, 
        na.strings=c("NA","NaN", " ") 
      )
    }
    else{ 
      data <- read.csv(
        paste(as.character(i), ".csv", sep=""),     ## Normal
        header = T, 
        na.strings=c("NA","NaN", " ") 
     )
    }
    ## getting rid of all the "NA" values and, consequently, all the non-complete ovservations (the ones with at least one NA in row)
    data = na.omit(data) 
    ## if the number of complete observed cases meets the quota, find the correlation between the pollutants for the given monitor AND
    ## store the results in the correlation matrix
    if (nrow(data) > threshold) {
      correlationVector = c(correlationVector, cor(data[,2], data[,3]))
    }
   }
  setwd("..")  # reseting working directory path
  return (correlationVector)
}