statistics using awk

Shell Programming and Scripting
1

Hi,

I have 3 columns in a file listed below.

X Y X/(X+Y)

1    1    0.5
1    1    0.5
4    1    0.8
1    1    0.5
6    1    0.857142857
1    1    0.5
23    1    0.958333333

Now I want to find confidence interval using the formula for each row.

(p-2 sqrt p(1-p)/(x+y), p+2 sqrt p(1-p)/(x+y)) =(a,b)

where p = x/(x+y) already calculated in column 3 of the file..

I want the values(a,b) to be placed in column 4.

Thanks,

Diya

2 
 nawk '{a=$3-2*sqrt($3*(1-$3)/($1+$2));b=$3+2 * sqrt($3*(1-$3)/($1+$2)); print $0, a","b}' myFile
3

Hi,

Thanks for the reply,

Firstly I cannot run using nawk as it says command not found.Probably I have to install it on my Linux machine.

When I ran the command with awk it generated me the following error.

awk '{a=$3-2*sqrt($3*(1-$3)/($1+$2));b=$3+2 * sqrt($3*(1-$3)/($1+$2)); print $0, a","b}' confi.txt
awk: (FILENAME=confi.txt FNR=1) fatal: division by zero attempted

Thanks,

4

given myFile:

1    1    0.5
1    1    0.5
4    1    0.8
1    1    0.5
6    1    0.857142857
1    1    0.5
23    1    0.958333333

the following:

nawk '{a=$3-2*sqrt($3*(1-$3)/($1+$2));b=$3+2 * sqrt($3*(1-$3)/($1+$2)); print $0, a","b}' myFile

produces:

1    1    0.5 -0.207107,1.20711
1    1    0.5 -0.207107,1.20711
4    1    0.8 0.442229,1.15777
1    1    0.5 -0.207107,1.20711
6    1    0.857142857 0.592623,1.12166
1    1    0.5 -0.207107,1.20711
23    1    0.958333333 0.876755,1.03991

What's your data file look like?
You probably need to cover the case where '$1+$2' results in '0'

5

Thanks.. It worked.. I was giving the header line so it did not work..

6

You might want to make the precedences of operators a little bit more explicit - based on YOUR understanding of the formulae:

a=($3-2)*sqrt($3*(1-$3)/($1+$2))
OR
a=$3-(2*sqrt($3*(1-$3)/($1+$2)))

and the same goes for the 'sqrt' and the others.....

7

Having Linux means you have GNU awk, meaning nawk isn't necessary. This is why it's always a good idea to say what your OS is when you ask a question...

8

Hi,

I have a file

 ABC  1001   1500
ABC 1200 1900
CDE 1700 2200
DEF 2100 3200
IJK  4100 4400
IJK  4500 4800

I want the file in such a way that if the name is same in column 1 (ABC) then it has to pick the longest length (column C -B) and output the file

o/p

ABC 1200  1900
CDE 1700 2200
DEF 2100 3200
IJK  4500 4800

How can i do it in unix or awk?

Thanks,

Diya