## Merging

Solved
Frequent Contributor
Posts: 130

# Merging

I am using the following code to merge data sets newl4 and newd4 after they have been sorted by wsubj:

data all;
merge newl4(in=inp) newd4(in=ine);
by wsubj ;
if inp and ine;
run;

newd4

subj time cmt conc

1 0.25 1 0
1 0.25 2 0
1 0.25 3 0
1 0.25 4 0
1 0.25 5 0
1 0.25 6 0
1 0.25 7 0
1 0.25 8 0
1 0.25 9 0
1 0.25 10 0
1 0.25 11 0.2242009
1 0.25 12 0
1 0.25 13 0
1 0.25 14 0
1 0.25 15 0
1 0.25 16 0
1 0.25 17 0
1 0.25 18 0
1 0.25 19 0
1 0.25 20 0
1 0.25 21 0
1 0.25 22 0
1 0.25 23 0
1 0.25 24 0

Newl4

subj time cmt conc

1 0.25 1 0
1 0.25 2 0
1 0.25 3 0
1 0.25 4 0
1 0.25 5 0
1 0.25 6 0
1 0.25 7 0
1 0.25 8 0
1 0.25 9 0
1 0.25 10 0
1 0.25 11 0
1 0.25 12 0
1 0.25 13 0
1 0.25 14 0
1 0.25 15 0
1 0.25 16 0
1 0.25 17 0
1 0.25 18 0
1 0.25 19 0
1 0.25 20 0
1 0.25 21 0
1 0.25 22 0
1 0.25 23 0.0960861

10.25 24 0

The only difference between them is that newd5 and newl4 have a conc in different cmt(i.e., newd4-cmt 11 and newl4 cmt23).

Whe I try to merge them I only get a value for conc in cmt 11 of 0.22, while cmt 23=0.  Can someone  give me code that will result in both cmt11 and cmt23 showing the values in newd4 and newl4 in the merged data set all?

Accepted Solutions
Solution
‎02-15-2017 05:33 AM
Posts: 5,485

## Re: Merging

If conc is always >= 0 in both datasets, then you can simply do:

``````data all;
merge newl4(in=inp) newd4(in=ine rename=conc=conc2);
by subj ;
if inp and ine;
conc = max(conc, conc2);
drop conc2;
run;``````
PG

All Replies
Posts: 4,697

## Re: Merging

[ Edited ]

Your variable conc exists in both datasets but is not part of the by statement for the merge. In such a case the result dataset will contain the value of conc of the last dataset where it exists (using the order as listed in the merge statement).

The easiest way to get around this: rename your variables so that the names are unique; ie. conc_newd4, conc_newl4

If you're after non-missing values from either dataset you then can use a coalesce() function in the data step where you merge, ie.

conc=coalesce(conc_newd4, conc_newl4);

You will still have to decide what to do in a case where there is a  non-missing value in the same row for both conc_newd4 and conc_newl4.

Solution
‎02-15-2017 05:33 AM
Posts: 5,485

## Re: Merging

If conc is always >= 0 in both datasets, then you can simply do:

``````data all;
merge newl4(in=inp) newd4(in=ine rename=conc=conc2);
by subj ;
if inp and ine;
conc = max(conc, conc2);
drop conc2;
run;``````
PG
Frequent Contributor
Posts: 130

## Re: Merging

The code worked very well.

Thanks

☑ This topic is solved.