Shell Programming and Scripting

Hi experts,

I want to group by average, for multiple columns starting column $7 until NF,
group by ($1-$5), please help

For just 7th column, I can do

Code:

awk '
    NR>1{
        arr[$1,$2,$3,$4,$5]   += $7
        count[$1,$2,$3,$4,$5] += 1
    }
    END{
        for (a in arr) {
            print a, arr[a]/count[a]
        }
    }
' FILE

you can use a for loop in this manner

Code:

 
 sum=0; for(I = 7; I <= NF; I++) { sum += $I; }

where $I would be values of your 7th, 8th etc... fields

Wouldn't that add all the elements in the same row? I need average for each of the columns separately.

Sorry, I misunderstood your request

Please show us some sample input, show us the output you expect from that sample input, tell us what OS and shell you're using, and tell us something about the ranges of bounds we should expect on the files you'll be processing (e.g., maximum number of fields, ranges of values in those fields, number of lines, is the number of fields within an input file a constant, etc.).

Does the code you showed us give you the results you want for column 7 in your input file?

sorry about not providing with a sample input, I`m using cygwin. The data range is 0 to 100,000 should be outputted upto 2 decimal places.
Data is 83000 lines, not very big.

Yes, the code calculates the average correctly for only the 7th column, although I should
populate arr as arr[$1" "$2" "$3" "$4" "$5] to get all the variables delimited.

Sample input has 4 data columns, I have many in the original data starting col7 until NF ($22).

Code:

  a1 b1 c1 d1 e1 12 13 14 15
  a1 b1 c1 d1 e1 14 15 16 17
  a1 b1 c1 d1 e1 13 14 15 16
  a2 b1 c1 d1 e1 112 113 114 115
  a2 b1 c1 d1 e1 114 115 116 117
  a2 b1 c1 d1 e1 113 114 115 116

Output should be

Code:

  a1 b1 c1 d1 e1 13 14 15 16
  a2 b1 c1 d1 e1 113 114 115 116

The code you showed us in your 1st post in this thread skips data in the 1st line of your file (which I assumed was intended to skip over a header line). But, I don't see any headers in this sample. Is there a header, or not? If there s a header, should it be copied to the output?

Is the number of fields constant in an input file, or can it vary from line to line?

It looks like there is a leading space in your sample input and output. Is a leading space required in your output?

Do you want 2 decimal places in all computed output fields, or do you want values to be printed without decimal places (as in your sample output) in cases where the computed result is an integral value?

You say you want to calculate averages for fields 7 through NF, but your sample data also calculates the average for field 6? Is field 6 supposed to be ignored in calculations and removed from the output, or is field 6 to be averaged as well as fields 7 through NF?