MAR Data Set Generation.R

roll_Die <- function(input_prob) {
  
  value <- rbinom(n=1,size=1,prob=input_prob)
  
  if (value == 1){
    print(TRUE)
    return(TRUE)
    
  }
  
  return(FALSE)
  
}

baseline_Check <- function(baseline, visit_Value) {
  
  if (visit_Value > baseline * 0.9) {
    case_prob = 0.3
  } else {
    case_prob = 0.05
  }
  return(case_prob)
}

clear_Values <- function(Subject_ID, visit_Number_Missing) {
  
  #compile columns (variables) to clear
  cols <- c()
  for (variable in c("AVAL", "CHG", "PCHG", "PCHGCAT1", 
                     "PCHGCA1N", "PCHGCAT2", "PCHGCA2N" )){
    col <- which(colnames(Short_DataSet) == variable)
    cols <- append(cols, col, after = length(cols))
  }
  
  #Compile rows (subjects/visits) to clear
  rows <- c()
  for (visit_Num in visit_Number_Missing:6) {
    row <- which(Short_DataSet$SUBJID == Subject_ID & Short_DataSet$AVISITN == visit_Num  )
    rows <- append(rows, row, after = length(rows))
  }
  
  #Set corresponding values to NA
  for(r in rows){
    for (c in cols){
      Short_DataSet[[c]][r] <- NA 
    
      
      }
    }
  Short_DataSet <<- Short_DataSet
}
  
generate_New_Test_Data <- function() {
  
    DataSet <- data.frame(read.csv("ADPA + SEX.csv"))
    
    # Trim bulk of data to simplify initial coding
    Subset_Index <<- sample(1:1831, 10, replace = FALSE) 
    
    Subset_Index <- sort(Subset_Index)
    
    Short_DataSet <<- DataSet[DataSet$SUBJID %in% c(Subset_Index), ]
    
    return()
    
}

load_Full_Data <- function() {
  
  DataSet <- data.frame(read.csv("ADPA + SEX.csv"))
  
  
  
  Subset_Index <<- c(1:1831)
  
  Short_DataSet <<- DataSet[DataSet$SUBJID %in% c(Subset_Index), ]
  
  return()
  
}

main <- function() {
  
  #generate_New_Test_Data()
  load_Full_Data()
  
  for (subjectID in Subset_Index){
    
    baseline <- Short_DataSet$AVAL[which(Short_DataSet$SUBJID == subjectID & Short_DataSet$ABLFL == "Y" & Short_DataSet$PARAMCD =="PASI" )]
    baseline <- as.integer(baseline)
    
    for (visit in 4:6) {
      
      visit_Value <- Short_DataSet$AVAL[which(Short_DataSet$SUBJID == subjectID & Short_DataSet$AVISITN == visit - 1 & Short_DataSet$PARAMCD =="PASI" )]
      
      
      
      prob <- baseline_Check(baseline, visit_Value)
      
      #print(subjectID)
      #print(visit)
      #print(prob)
      
      missing <- roll_Die(prob)
      
      #print(missing)
      
      if (missing) {
        
        clear_Values(subjectID, visit)
        
        break
        
      }
        
      
      
    }
    
  }
}

##############################################################

main()

write.csv(Short_DataSet,"C:\\Users\\Charles Watt\\Desktop\\UConn Biostats\\Biostats Practicum\\MAR_Data_V1.csv", row.names = FALSE)

#Hard-coded testing commands

#Short_DataSet <- clear_Values(167, 6)

#baseline_Check(35, 31)