#TJHSST ~ Simple Statistics

statistics.c

/*
  Statistics
*/
#include <stdio.h>
#include <math.h>
#include "statisti.h"

main(argc,argv)
int argc;
char *argv[];
{
  FILE *datafile;
  double x,y,sumx,sumy,sumxy,sumx2,sumy2,stddevx,stddevy,r,b,a,xbar,ybar;
  int n;

  if(argc<2)  {
    fputs("Enter Data Filename\n",stderr);
    exit(1);
  }
  datafile=fopen(argv[1],"r");
  if(!datafile)  {
    fputs("Error Opening Data File\n",stderr);
    exit(1);
  }
  sumx=sumy=sumxy=sumx2=sumy2=0.0;
  n=0;
  while(fscanf(datafile,"%lf%*c%lf%*c",&x,&y)!=EOF)  {
    sumx+=x;
    sumy+=y;
    sumxy+=(x*y);
    sumx2+=(x*x);
    sumy2+=(y*y);
    n++;
  }
  fclose(datafile);
  xbar=sumx/n;
  ybar=sumy/n;
  stddevx=stat_stddev(sumx,sumx2,n);
  stddevy=stat_stddev(sumy,sumy2,n);
  r=linear_r(sumxy,sumx,sumy,stddevx,stddevy,n);
  b=linear_b(r,stddevx,stddevy);
  a=linear_a(xbar,ybar,b);
  printf("Data points : %3d\n",n);
  printf("Means\n");
  printf("    x bar   : %3f\n",xbar);
  printf("    y bar   : %3f\n",ybar);
  printf("    %c x     : %3f\n",(char)228,sumx);
  printf("    %c y     : %3f\n",(char)228,sumy);
  printf("    %c xy    : %3f\n",(char)228,sumxy);
  printf("    %c x^2   : %3f\n",(char)228,sumx2);
  printf("    %c y^2   : %3f\n",(char)228,sumy2);
  printf("Standard Deviations\n");
  printf("    sx     : %3f\n",stddevx);
  printf("    sy     : %3f\n",stddevy);
  printf(" y = a + bx\n");
  printf("     a      : %3f\n",a);
  printf("     b      : %3f\n",b);
  printf("Correlation Coefficent\n");
  printf("     r      : %3f\n",r);
}

double stat_stddev(sumx,sumx2,n)
double sumx,sumx2;
int n;
{
  if(n<=1)
    return (-1.0);
  return (1/(n-1.0))*((sumx2)-(sumx*sumx)/n);
}

double linear_r(sumxy,sumx,sumy,stddevx,stddevy,n)
double sumxy,sumx,sumy,stddevx,stddevy;
int n;
{
  if(n<=1)
    return (-1.0);
  return (sumxy-(sumx*sumy/n))/((n-1)*stddevx*stddevy);
}

double linear_b(sumxy,sumx,sumy,sumx2,n)
double sumxy,sumx,sumy,sumx2;
int n;
{
  return (r*stddevx/stddevy);
}

double linear_a(xbar,ybar,b)
double xbar,ybar,b;
{
  return (ybar-b*xbar);
}

A C program found on an old 5.25 floppy disk from when I was a nerd at Thomas Jefferson High School for Science and Technology between 1988 and 1992.

The program calculates (and displays) simple statistics of a bi-variate data set.

Anyone can do whatever they'd like to with this program--if anything.

I remain a math nerd.