[Gretl-users] Re: "join"two gdt panel data files (B) save them as time series

Am 07.06.2022 um 18:28 schrieb Ioannis Venetis: Hi Ioannis, Actually it seems to be the other way around. I think the trick here would be to first extend the time dimension of the earlier dataset (example2) to cover everything up to 2022, being empty until the actual appending is done. (menu Data/Add observations) Simple appending still doesn't seem to work then (maybe it should), but you can use the graphical join option, then in the dialog window specify "TIME" and "GEO" as the keys to be matched. In the GUI right now you have to do that separately for each variable, I'm afraid. (In a script it's easier for multiple variables, but here you only have two.) Hope that explanation was clear. Ah, that's a classic - well, I just mean that I've had this type of problem myself. I don't think there's a super easy way. One solution is to convert the stuff to a matrix, reshape that, and then reload it into a clean time series dataset. The key is to know that gretl's panel storage is as stacked time series, so a series becomes an NT,1 vector (when missings are included). Something like: <hansl> set skip_missing off matrix M00 = {CP00} N = max($unit) T = max($obsminor) # $obsminor is time in panels M00 = mshape(M00, T, N) </hansl> And then you would perhaps store the matrix M00, prepare a time series dataset of length T, and load M00 and turn it into your series collection. (Perhaps using the mat2list function.) HTH, sven