Thank you very much! I hoped there could have been a more direct way, but as you said since there is no boolean column type I guess there is no other way but manual. ... View more

Thanks! I adapted to work in CAS, and yes it works:   data P.want / single=yes; set P.have; by ID; retain aa bb; if a then aa = a; else a = aa; if b then bb = b; else b = bb; drop aa bb; run; However if anyone knows if there are dedicated actions for these situations in Viya 3.5 or planned for Viya 4, that would be much appreciated! ... View more

Unfortunately that code doesn't work in CAS:   cas session; libname P cas caslib="PUBLIC"; data P.have; input ID A B; datalines; 1 10 20 2 . 21 3 . . 4 11 . ; data P.temp / view = P.temp; set P.have; i = 1; run; data P.want (drop = i); update P.temp(obs=0) temp; by i; output; run; 95 data P.temp / view = P.temp; 96 set P.have; 97 i = 1; 98 run; ERROR: P.TEMP.VIEW cannot be opened because files of type VIEW are not supported in the P library. ERROR: Unable to save DATA STEP view P.TEMP. ... View more

Yes I can, but it requires a little bit of coding with lags and conditions, I was wondering whether it would exists an action to quickly handle these classic use cases (I miss proc expand and proc timeseries a lot in CAS). ... View more

Thanks it works great. I'll flag as solution the previous answer since this is a collateral issue CAS related. Regards ... View more

Same as the previous, but with a CAS library (instead of work):   cas session; libname CASUSER cas caslib="CASUSER"; data CASUSER.HAVE; format timestamp datetime25.; timestamp=datetime(); A=1; B=500000; C=0.03; output; timestamp=datetime()+60; A=2; B=100000; C=0.05; output; timestamp=datetime()+120; A=3; B=300000; C=0.01; output; timestamp=datetime()+180; A=4; B=400000; C=0.02; output; timestamp=datetime()+240; A=5; B=200000; C=0.04; output; run; proc transpose data=CASUSER.Have out=CASUSER.Want name=Variable; by timestamp; var A B C; run; ods graphics / reset; /* plot on a single scale in multiple graphs */ proc sgpanel data=CASUSER.Want; panelby Variable / UNISCALE=COLUMN columns=1; series x=timestamp y=Col1; run; ERROR: One or more ID variables are required. ERROR: Failure during action initialization. ERROR: The action stopped due to errors. NOTE: The Cloud Analytic Services server processed the request in 0.046021 seconds. NOTE: The SAS System stopped processing this step because of errors. NOTE: PROCEDURE TRANSPOSE ha utilizzato (tempo totale di elaborazione): real time 0.15 seconds cpu time 0.05 seconds ... View more

Thanks that's exactly what I needed! However I'm getting this error for proc transpose: ERROR: One or more ID variables are required. ERROR: Failure during action initialization. ERROR: The action stopped due to errors. since the table is in CAS. In SAS 9.4 it works, but in CAS it says the id parameter is required. What's wrong? ... View more

Maybe it's clear in my head but I'm not succeeding in explain it 😄   However no, in my example the x axis values are exactly the same for all three graphs, so setting min and max wouldn't change anything since the min and the max are already reached by all three graphs.   What I mean is that since the three vars have different scale (and that's ok), when the sgplot draws the y axis, it is drawn "more to the right" depending on what it has to plot on the left of the y axis, that is the numbers of the scale.   In the first graph, the numbers on the left are "1" "2" "3",... assume they occupy 10px, then there is a white space, assume it occupies 5px, and then the y axis: so it begins in the paper at coordinate 15px.   In the second graph, the numbers in the left are "100000" "200000" etc, so they occupy MORE space, assume 10px * 6 digits = 60px, then there is a white space, assume it occupies 5px as before, and then the y axis: so it begins in the paper at coordinate 65px.   Hence the two y axes are not aligned vertically! One is at 15px, one is at 65px.   My question is how to "center" the graphs so that the y axes are exactly at the same distance in pixel from the left of the paper.  ... View more

No, I don't want to use the same scale for all graphs, every one should have its own scale. But if I look at the y axes vertically, they should be aligned. Let me attach an image to explain what I mean: ... View more

Sure, since the dataset is very simple I assumed it could be discussed just knowing the dataset metadata!   Here's the dataset and my code so far: data HAVE; format timestamp datetime25.; timestamp=datetime(); A=1; B=500000; C=0.03; output; timestamp=datetime()+60; A=2; B=100000; C=0.05; output; timestamp=datetime()+120; A=3; B=300000; C=0.01; output; timestamp=datetime()+180; A=4; B=400000; C=0.02; output; timestamp=datetime()+240; A=5; B=200000; C=0.04; output; run; proc sgplot data=have; series x=timestamp y=A; run; proc sgplot data=have; series x=timestamp y=B; run; proc sgplot data=have; series x=timestamp y=C; run;   As you can see the three graphs are not vertically aligned on y axes. Moreover, I'd prefer not to create an ad-hoc temporary table, since it would require more time (in real data rows are millions and columns A B C are some dozens).   Thanks ... View more

Thanks I taught about it, however I couldn't figure out how to use it in this case, there is no groups to use in panelby and I don't want to restructure the dataset if not absolutely necessary. Can you give me an example of how should I use sgpanel here? Thanks ... View more