I have a data set with three columns. One is series of times. The second is a column which is mainly missing values. The third is also numeric.
What I need to do is, for every non-missing value in the second column, create a data set with the mean value of the third column for every observation before the time of the non-missing value that is within three hours of that observation.