data/scripts/fill_NAs.R

   1 fill_NAs = function(file)
   2 {
   3     dat_with_date = read.csv(file)
   4     dat = dat_with_date[,-1] #get rid of date column
   5
   6     line = 1
   7     n = nrow(dat)
   8     p = ncol(dat)
   9     last_noNA_indices = rep(0, p)
  10     while (line <= n)
  11     {
  12 #       print(line)
  13         for (i in 1:p)
  14         {
  15 #           if (is.numeric(dat[line,i]))
  16             if (!is.na(dat[line,i]))
  17             {
  18                 if (last_noNA_indices[i] < line-1)
  19                 {
  20                     #do some completion
  21                     meanVal = ifelse(last_noNA_indices[i]>0,
  22                                                  0.5*(dat[last_noNA_indices[i],i]+dat[line,i]),
  23                                                  dat[line,i])
  24
  25                     for (j in (last_noNA_indices[i]+1):(line-1))
  26                         dat[j,i] = meanVal
  27                 }
  28                 last_noNA_indices[i] = line
  29             }
  30         }
  31         line = line + 1
  32     }
  33
  34     #complete until end if needed
  35     for (i in 1:p)
  36     {
  37         if (last_noNA_indices[i] < n)
  38         {
  39             for (j in (last_noNA_indices[i]+1):n)
  40                 dat[j,i] = dat[last_noNA_indices[i],i]
  41         }
  42     }
  43
  44     dat_with_date[,2:(p+1)] = dat
  45     write.csv(dat_with_date, "NONA.csv", row.names=FALSE)
  46 }