Re: [Gretl-users] Question about join

On Tue, 9 Dec 2014, Logan Kelly wrote: Well, not really. Simply, "join" is not set up for adding time-series data of lower frequency to an existing dataset of higher frequency automatically. To use join for this purpose you'd have to supply some specific key variables. For example: <hansl> nulldata 24 setobs 12 2012:01 series year = $obsmajor series quarter = 1+floor(($obsminor-1)/3) </hansl> Then if you have a data file that looks like this year, quarter, ENU2705340010 2013, 4, 1209 2013, 3, 1163 2013, 2, 1142 2013, 1, 1276 2012, 4, 1239 2012, 3, 1135 2012, 2, 1123 2012, 1, 1278 you can do: join filename.csv out --ikey=year,quarter --data="ENU2705340010" This will write the given quarter's data into each month of the quarter. An alternative approach is to read the quarterly data into gretl first, expand the data using "dataset expand" (with either repetition of values or interpolation), store the expanded (now monthly) data, then use join. Note that in your second example above, --aggr=avg is not going to do anything. The --aggr option applies when n > 1 values from the outer data file match a single row on the left. Your case is one where a single row in the outer data file matches more than one slot on the left. Join provides no special option for dealing with this case, it just copies the right-hand value into all matching locations on the left. Allin Cottrell