| 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 | for (i in 1:p) |
| 13 | { |
| 14 | if (!is.na(dat[line,i])) |
| 15 | { |
| 16 | if (last_noNA_indices[i] < line-1) |
| 17 | { |
| 18 | #do some completion |
| 19 | meanVal = ifelse( last_noNA_indices[i]>0, |
| 20 | 0.5*(dat[last_noNA_indices[i],i]+dat[line,i]), |
| 21 | dat[line,i] ) |
| 22 | dat[(last_noNA_indices[i]+1):(line-1),i] = meanVal |
| 23 | } |
| 24 | last_noNA_indices[i] = line |
| 25 | } |
| 26 | } |
| 27 | line = line + 1 |
| 28 | } |
| 29 | |
| 30 | #complete until end if needed |
| 31 | for (i in 1:p) |
| 32 | { |
| 33 | if (last_noNA_indices[i] < n) |
| 34 | dat[(last_noNA_indices[i]+1):n,i] = dat[last_noNA_indices[i],i] |
| 35 | } |
| 36 | |
| 37 | dat_with_date[,2:(p+1)] = dat |
| 38 | write.csv(dat_with_date, "NONA.csv", row.names=FALSE) |
| 39 | } |