I don't understand the problem yet. But a simple regression with 100*10000 and 10 regressor takes about 2.12 seconds as showed below. I don't this time is a problem here.
47 data t1;
48 array x(10) ;
49 do i=1 to 100*10000;
50 do j=1 to 10;
51 x[j]=rannor(123);
52 end;
53 y=1+sum(of x[*]) + rannor(123);
54 output;
55 end;
56 keep y x:;
57 run;
NOTE: The data set WORK.T1 has 1000000 observations and 11 variables.
NOTE: DATA statement used (Total process time):
real time 1.39 seconds
cpu time 1.40 seconds
58
59 proc reg data=t1;
60 model y=x1-x10;
61 run;
62 quit;
NOTE: PROCEDURE REG used (Total process time):
real time 5.70 seconds
cpu time 2.12 seconds
|