Remove NA when using "order"

Question

I have this huge matrix of data with columns for year, month, day and precipitation which I need to order and also delete the row when the precipitation is NA (which happens on the day 31 of every month that only has 30 days and Februaries...). After consulting the r help files I used the following code:

dat<- aa[order(aa$year, aa$month, aa$day, na.last=NA),] 

It ordered my data perfectly but I still have all the NAs... Can anyone tell me why it hasn't work?

thanks

> head(dat)
    code year month                  station ALTITUD   PROV LONGITUD LATITUD day P1 id
1.1 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   1  0  1
1.2 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   2  0  1
1.3 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   3  0  1
1.4 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   4  0  1
1.5 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   5  0  1
1.6 3059 1940    11 ALBALATE DE LAS NOGUERAS     855 CUENCA   216372  402200   6  0  1

Answer 1

The na.last argument to order only removes NA from the objects passed to order via .... Your NA are in aa$precipitation, not aa$year, aa$month, or aa$day, so you need:

dat <- na.omit(aa[order(aa$year, aa$month, aa$day),])

You may want to consider using a time-series class like zoo or xts for time-series data.