Computing Averages for values in different columns and different observations

[Sean100]

Hi all,

I have a table like this:

A                B

-63261.57  -0.76
-0.75          0
0.01           68.63
68.64         252.5
252.51       488.93
488.94       787.53
787.54       1114.01
1114.02     1635.63
1635.64     2619.87
2619.91     64905.18

and I want to get the average of the record in the second column and the record in the following observation of the first column.
So for example, the average of -0.76 (row 1 column 2) and -0.75 (row 2 column 1), and so on. 

Any ideas on how to go about this? Also, I will not necessarily have 10 observations every time.

Thanks a million! 

[RW9]

Maybe something like (and not tested, present test data in the form of a datastep):

data want;
  set have;
  if _n_ ne 1 then result=avg(lag(b),a);
run;

[Sean100]

Hi RW9,

This is working almost perfectly, but it doesn't calculate the very first possible entry. From the example I gave I get the following:

A                  B              Result

-63261.57    -0.76         .
-0.75            0               .                <-Here is the problem
0.01             68.63         0.005
68.64           252.5         68.635
252.51         488.93       252.505
488.94         787.53       488.935
787.54         1114.01     787.535
1114.02       1635.63     1114.015
1635.64       2619.87     1635.635
2619.91       64905.18   2619.89

Any idea about why this is happening/ how can I fix it?

Thanks

[RW9 · Accepted solution]

Yes, minor change:

data want;
  set have;
  if _n_ > 0 then result=avg(lag(b),a);
run;

[Astounding]

Since _n_ will always be greater than zero, you could use:

result = avg(lag(b), a);

if _n_=1 then result = .;

[Sean100]

Brilliant, thanks!