6.3 Class LeastSqFit

Class LeastSqFit can be used to calculate linear regressions with or without weighting factors. Once constructed, a LeastSqFit can be interrogated for various properties of the fit. Here's an example:

#include <rw/lsqfit.h>
#include <rw/math/mathvec.h>
#include <rw/rand.h>
#include <iostream.h>

main()
{
  // Two straight lines:
  RWMathVec<double> x(50, 0, 1);
  RWMathVec<double> y(50, 0, 1);

  // Random number generator:
  RWTRandUniform<RWRandGenerator> ru;

  // Add some noise to y:
  y += ru.randValue(50);

  // Construct the fit, without using weighting factors:
  LeastSqFit lsf(x, y);

  // Now get various properties of the fit:
  cout << "Correlation coefficient is " <<
                                  lsf.correlationCoeff() << endl;
  cout << "The slope of the fit is "<<
                                  lsf.slope() << endl;