linregress.cwp.lib

su 的源代码库

LINEAR_REGRESSION - Compute linear regression of (y1,y2,...,yn) against 
(x1,x2,...,xn)
Function Prototype:
void linear_regression(float *y, float *x, int n, float coeff[4]);
Input:
y		array of y values
x		array of x values
n		length of y and x arrays
Output:
coeff[4] where:

coeff[0]	slope of best fit line
coeff[1]	intercept of best fit line
coeff[2]	correlation coefficient of fit (1 = perfect) [dimensionless]
coeff[3]	standard error of fit (0 = perfect) [dimensions of y]
Notes: 

y(x) 
    |      *  .    fit is  y(x) = a x + b
    |       .          
    |     .  *
    | * .    
    | . *         
     ------------------- x
     
         n Sum[x*y] - Sum[x]*Sum[y]
     a = --------------------------
         n Sum[x*x] - Sum[x]*Sum[x]
         
         Sum[y] - a*Sum[x]
     b = -----------------
                n
    
     cc = std definition
     
     se = std definition
    
Author:  Chris Liner, UTulsa, 11/16/03