AWK script to split data and find average

Shell Programming and Scripting
1

Input:

2.58359023380340e+02
1.43758864405595e+02
-7.65700666212508e+00
1.06460208083228e+02
1.26185441783936e+02
-3.41389169427027e+01
-1.40393299309592e+02
-3.07758776849508e+01
1.45067703495838e+02
1.79405834959073e+02
5.06666234594205e+01

OUT

2.0105894389e+02   (average of record 1 +2)
1.1632282493e+02  (average of record 3+4)
etc.

I need a script that takes every 2 records and averages them out,
so a file with 50,000 records would be converted to a file of 25,000 records.

I tried with this script but it is not working:

(NR%2)!=0 {
        for (i=1; i<=NF; i++){
                sum+=$i
        }
}
(NR%2)==0 {
        for (i=1; i<=NF; i++) printf "%f ", sum/2
        printf "\n"
        delete sum
}
2

Seems to work okay...

$ cat file1
2.58359023380340e+02
1.43758864405595e+02
-7.65700666212508e+00
1.06460208083228e+02
1.26185441783936e+02
-3.41389169427027e+01
-1.40393299309592e+02
-3.07758776849508e+01
1.45067703495838e+02
1.79405834959073e+02
5.06666234594205e+01

$ awk '
> (NR%2)!=0 {
>         for (i=1; i<=NF; i++){
>                 sum+=$i
>         }
> }
> (NR%2)==0 {
>         for (i=1; i<=NF; i++) printf "%f ", sum/2
>         printf "\n"
>         delete sum
> }
> ' file1
129.179512
-3.828503
63.092721
-70.196650
72.533852

$

What did you get?

3 
awk '{sum+=$1}!(NR%2){print sum/2;sum=0}END{if(NR%2) print sum}' OFMT='%.10e' file

giving

2.0105894389e+02
4.9401600711e+01
4.6023262421e+01
-8.5584588497e+01
1.6223676923e+02
5.0666623459e+01

for your sample.

1 Like
4

You forgot to add the numbers of the even lines. The code sould be :

sum/2 ==> (sum + $i)/2
1 Like
5

Thanks, elixir, this works it seems.

How would I change the script so if there is an odd number of values, the last record is discarded?

6

Just drop the END pattern and the corresponding action:

awk '{sum+=$1}!(NR%2){print sum/2;sum=0}' OFMT='%.10e' file
7

Ok, but this script removes the final record even for files with an even number of records.

8

for even records Elixir's solution works perfect..

For odd records - It will print last column.
Try with this..

 awk '{sum+=$1}!(NR%2){print sum/2;sum=0}END{if(sum != 0) {print sum}}' OFMT='%.10e' file
9 
$ awk '{m=$1;getline;sum=(m+$1)/2;print sum;}' OFMT='%.10e' new
OUTPUT:
2.0105894389e+02
4.9401600711e+01
4.6023262421e+01
-8.5584588497e+01
1.6223676923e+02
5.0666623459e+01