This is a good application for the MODIFY statement, in which observations are deleted, added, or replaced in place. I.e. the whole dataset is not copied to a new version just to change a very few records.
The following code will take a single record from a transaction data set (i.e. your new data each Monday), and if it does not have the same date as the end of the master file, the record is appended. But if it has the same date, then the last master record is merely re-written with the updated values:
data master;
input date mmddyy10. Calls_Made Calls_Cancelled;
format date date9.;
datalines;
05/01/2018 5000 2000
06/01/2018 5800 2300
07/01/2018 7600 2100
08/01/2018 6600 1000
09/01/2018 5600 3400
10/01/2018 4400 2500
run;
data transaction;
input date mmddyy10. calls_made calls_cancelled;
format date date9.;
datalines;
11/01/2018 3400 2400
run;
data master;
modify master nobs=nmaster point=nmaster;
set transaction (rename=(date=tran_date));
if date =tran_date then replace;
else do;
date=tran_date;
output;
end;
run;
The first time you run the above, a November record is added. If the following Monday, you make a new transaction dataset with DATE=11/01/2018 and new values for calls_made, calls_cancelled. and then run the last data step above, you'll replace the values for November.
Note the MODIFY statement reads only the last record in dataset master, no matter how many observations are in master. Sequential reading is avoided.
