Hi, I am new to IML, I am trying to add the reverse diagonals of each column

[Xamius32]

For example, Column 1 Row 1 would be a total. The Column 2 Row 1 plus Column 1 Row 2, then COL3 Row1+Col2 Row 2+ Col1 Row3, etc. For however many columns I have. It isnt a square matrix so once the columns exceed the rows it wouldnt match to the first column.

Header 1	Header 2	Header 3	Header 4	Header 5
	A	B	C	D
A	B	C	D
B	C	D

So for example above, I would want the totals of A, B, C, D etc.

I have the matrix, now just not sure how to add up what I need.

Any ideas?

[Xamius32]

Thinking about this more, it could be easier if I could figure out how to shift over rows by X amount. So for row 2, I would shift it over by 1, and the last observation would be deleted. Is that possible?

[IanWakeling]

I have a function to do exactly this.   It creates the 'shift' that you mention by manipulating the row index numbers of a matrix in its sparse format (details of the format can be found in the documentation for the SPARSE function).  The result when converted back to normal form, with the FULL function, is a new matrix with one row for every antidiagonal.  Finally it is a simple matter to sum the rows of this new matrix.

start AntiDiagSum( x );
  nr = nrow( x );
  nc = ncol( x );
  s = sparse( x );
  s[ , 2] = s[ , 2] + s[ , 3] - 1;
  return( full( s , nr + nc - 1, nc )[ , +] );
finish;

/* get example 3x5 matrix */
a = shape(1:15, 3, 5);
print a;

/* required sums are from the 2nd to 5th antidiagonals */
b = AntiDiagSum(a)[2:5];
print b;

[Ksharp]

I don't how to do it with IML . But I can get it via data step. Maybe Rick can give IML solution.

data have;
input a1-a5;
cards;
99 1 2 3 4
1 2 3 4 99
2 3 4 99 99
;
run;
data temp;
 set have;
 array x{*} a1-a5;
  do j=dim(x)-_n_+1 to 1 by -1;
   x{j+_n_-1}=x{j};
  end;
  do i=1 to _n_-1;
   x{i}=.;
  end;
drop i j;
run;
proc means data=temp sum;
var a2-a5;
run;

Xia Keshan