Program that reads data from a file and calculates the mean?

[youngfreedman]

Hi everyone.

I'm trying to write a program that reads data from 2 files and then calculates the mean, standard deviation and standard error of both files (separate values for each). I'm struggling to get my head around simple I/O, so excuse the poor attempt, but this is what I have so far: (I'm only attempting to just print out each value for now.)

    program data
    implicit none

    integer             :: j
    double precision    :: test

    open(unit = 100, file = 'tmax_1910.txt', status = 'old', action = 'read')
     do j = 1,  12
     read(100,*) test
     print *, 'N1=', test
    end do

    end program data

If it helps, the file is a list of monthly rainfalls for a year.

Thanks for any help!

 

[FactChecker]

I can't immediately see anything wrong. Does it compile, link, execute? If you get an error message in one of those steps, that is the time to ask questions. And tell us what the error message is complete with line number.

 

[youngfreedman]

I can't immediately see anything wrong. Does it compile, link, execute? If you get an error message in one of those steps, that is the time to ask questions. And tell us what the error message is complete with line number.

I've actually made a lot of progress since posting this, I was going to delete the post but don't know how. However, I do still need help. Here's my code at this point:

    program data
    implicit none

    integer             :: R, F
    double precision    :: x, sum = 0, mean, y, mean2, sum2 = 0, var =0, sdv,
    var2 = 0, sdv2    open(unit = 100, file = 'tmax_1910.txt', status = 'old', action = 'read')
     do R = 1,  12
     read(100,*) x
     sum = sum + x

    end do

    mean = (sum)/12

     do R = 1, 12
      var = var + (((x - mean)**2.0)/12)
      sdv = var**0.5

    end do
    open(unit = 200, file = 'tmax_2010.txt', status = 'old', action = 'read')
     do F = 1, 12
     read(200,*) y
     sum2 = sum2 + y

    end do

    mean2 = (sum2)/12

     do F = 1, 12
      var2 = var2 + (((y - mean2)**2.0)/12)
      sdv2 = var2**0.5
    end do

    print *, 'mean=', mean, 'mean2=', mean2, 'sdv=', sdv, 'sdv2=', sdv2

    end program data

This prints the correct mean values, but the values for standard deviation (for both files) is incorrect. For reference, here are the numbers in each file:

File 1: 5.0, 6.6, 9.3, 10.4, 14.0, 18.0, 16.9, 18.6, 15.4, 13.1, 5.4, 7.6 (actual standard dev = 4.9..., my value 4.09..)
File 2:3.2, 4.3, 9.5, 13.0, 14.5, 19.2, 20.8, 19.0, 17.2, 12.9, 7.2, 2.0. (actual standard dev = 6.6.., my value 9.9...)

Thanks.

EDIT: Just to add, the values I get for mean are correct.

 

[FactChecker]

Your values of x are changing correctly in the first loop but once you get out of that loop, you are left with x = last value read. So the second loop has a constant x value. Your alternatives are to loop through the x values twice (either saving an array of them or reading them twice) or using a different formula for the variance.

You can use a formula for variance where you accumulate the sum of x² in the first loop and the sample mean. That saves you from looping through the x values twice.

 

[youngfreedman]

Your values of x are changing correctly in the first loop but once you get out of that loop, you are left with x = last value read. So the second loop has a constant x value. Your alternatives are to loop through the x values twice (either saving an array of them or reading them twice) or using a different formula for the variance.

You can use a formula for variance where you accumulate the sum of x² in the first loop and the sample mean. That saves you from looping through the x values twice.

I hadn't considered that. That makes sense, thanks !

 

[jim mcnamara]

The algorithm @FactChecker mentions is called an online algorithm. The idea is that you can type a single datastream at the keyboard and do things with the data like calculate mean, standard deviation, and variance. Works for a file, too: you just read through the file one time.

Wikpedia has examples if you search for 'variance', the original Knuth version is easy to understand.