LinearRegression.h

Go to the documentation of this file.

// --------------------------------------------------------------------------
//                   OpenMS -- Open-Source Mass Spectrometry
// --------------------------------------------------------------------------
// Copyright The OpenMS Team -- Eberhard Karls University Tuebingen,
// ETH Zurich, and Freie Universitaet Berlin 2002-2021.
//
// This software is released under a three-clause BSD license:
//  * Redistributions of source code must retain the above copyright
//    notice, this list of conditions and the following disclaimer.
//  * Redistributions in binary form must reproduce the above copyright
//    notice, this list of conditions and the following disclaimer in the
//    documentation and/or other materials provided with the distribution.
//  * Neither the name of any author or any participating institution
//    may be used to endorse or promote products derived from this software
//    without specific prior written permission.
// For a full list of authors, refer to the file AUTHORS.
// --------------------------------------------------------------------------
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
// ARE DISCLAIMED. IN NO EVENT SHALL ANY OF THE AUTHORS OR THE CONTRIBUTING
// INSTITUTIONS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// --------------------------------------------------------------------------
// $Maintainer: Timo Sachsenberg $
// $Authors: $
// --------------------------------------------------------------------------
 
#pragma once
 
#include <OpenMS/CONCEPT/Types.h>
#include <OpenMS/CONCEPT/Exception.h>
#include <OpenMS/DATASTRUCTURES/String.h>
#include <OpenMS/MATH/STATISTICS/RegressionUtils.h>
 
#include "Wm5Vector2.h"
#include "Wm5ApprLineFit2.h"
#include "Wm5LinearSystem.h"
 
#include <cmath>
#include <vector>
 
namespace OpenMS
{
  namespace Math
  {
    class OPENMS_DLLAPI LinearRegression
    {
public:
 
      LinearRegression() :
        intercept_(0),
        slope_(0),
        x_intercept_(0),
        lower_(0),
        upper_(0),
        t_star_(0),
        r_squared_(0),
        stand_dev_residuals_(0),
        mean_residuals_(0),
        stand_error_slope_(0),
        chi_squared_(0),
        rsd_(0)
      {
      }
 
      virtual ~LinearRegression() = default;
 
      template <typename Iterator>
      void computeRegression(double confidence_interval_P, Iterator x_begin, Iterator x_end, Iterator y_begin, bool compute_goodness = true);
 
      template <typename Iterator>
      void computeRegressionWeighted(double confidence_interval_P, Iterator x_begin, Iterator x_end, Iterator y_begin, Iterator w_begin, bool compute_goodness = true);
 
      double getIntercept() const;
      double getSlope() const;
      double getXIntercept() const;
      double getLower() const;
      double getUpper() const;
      double getTValue() const;
      double getRSquared() const;
      double getStandDevRes() const;
      double getMeanRes() const;
      double getStandErrSlope() const;
      double getChiSquared() const;
      double getRSD() const;
 
      static inline double computePointY(double x, double slope, double intercept)
      {
        return slope * x + intercept;
      }
 
protected:
 
      double intercept_;
      double slope_;
      double x_intercept_;
      double lower_;
      double upper_;
      double t_star_;
      double r_squared_;
      double stand_dev_residuals_;
      double mean_residuals_;
      double stand_error_slope_;
      double chi_squared_;
      double rsd_;
 
 
      void computeGoodness_(const std::vector<Wm5::Vector2d>& points, double confidence_interval_P);
 
      template <typename Iterator>
      double computeChiSquare(Iterator x_begin, Iterator x_end, Iterator y_begin, double slope, double intercept);
 
      template <typename Iterator>
      double computeWeightedChiSquare(Iterator x_begin, Iterator x_end, Iterator y_begin, Iterator w_begin, double slope, double intercept);
 
private:
 
      LinearRegression(const LinearRegression& arg);
      LinearRegression& operator=(const LinearRegression& arg);
 
    }; //class
 
    //x, y, w must be of same size
    template <typename Iterator>
    double LinearRegression::computeChiSquare(Iterator x_begin, Iterator x_end, Iterator y_begin, double slope, double intercept)
    {
      double chi_squared = 0.0;
      Iterator xIter = x_begin;
      Iterator yIter = y_begin;
      for (; xIter != x_end; ++xIter, ++yIter)
      {
        chi_squared += std::pow(*yIter - computePointY(*xIter, slope, intercept), 2);
      }
 
      return chi_squared;
    }
 
    //x, y, w must be of same size
    template <typename Iterator>
    double LinearRegression::computeWeightedChiSquare(Iterator x_begin, Iterator x_end, Iterator y_begin, Iterator w_begin, double slope, double intercept)
    {
      double chi_squared = 0.0;
      Iterator xIter = x_begin;
      Iterator yIter = y_begin;
      Iterator wIter = w_begin;
      for (; xIter != x_end; ++xIter, ++yIter, ++wIter)
      {
        chi_squared += *wIter * std::pow(*yIter - computePointY(*xIter, slope, intercept), 2);
      }
 
      return chi_squared;
    }
 
    template <typename Iterator>
    void LinearRegression::computeRegression(double confidence_interval_P, Iterator x_begin, Iterator x_end, Iterator y_begin, bool compute_goodness)
    {
      std::vector<Wm5::Vector2d> points = iteratorRange2Wm5Vectors(x_begin, x_end, y_begin);
 
      // Compute the unweighted linear fit.
      // Get the intercept and the slope of the regression Y_hat=intercept_+slope_*X
      // and the value of Chi squared (sum( (y - evel(x))^2)
      bool pass = Wm5::HeightLineFit2<double>(static_cast<int>(points.size()), &points.front(), slope_, intercept_);
      chi_squared_ = computeChiSquare(x_begin, x_end, y_begin, slope_, intercept_);
 
      if (pass)
      {
        if (compute_goodness && points.size() > 2) computeGoodness_(points, confidence_interval_P);
      }
      else
      {
        throw Exception::UnableToFit(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION,
            "UnableToFit-LinearRegression", String("Could not fit a linear model to the data (") + points.size() + " points).");
      }
    }
 
    template <typename Iterator>
    void LinearRegression::computeRegressionWeighted(double confidence_interval_P, Iterator x_begin, Iterator x_end, Iterator y_begin, Iterator w_begin, bool compute_goodness)
    {
      // Compute the weighted linear fit.
      // Get the intercept and the slope of the regression Y_hat=intercept_+slope_*X
      // and the value of Chi squared, the covariances of the intercept and the slope
      std::vector<Wm5::Vector2d> points = iteratorRange2Wm5Vectors(x_begin, x_end, y_begin);
      // Compute sums for linear system. copy&paste from GeometricTools Wm5ApprLineFit2.cpp
      // and modified to allow weights
      int numPoints = static_cast<int>(points.size());
      double sumX = 0, sumY = 0;
      double sumXX = 0, sumXY = 0;
      double sumW = 0;
      Iterator wIter = w_begin;
 
      for (int i = 0; i < numPoints; ++i, ++wIter)
      {
        sumX += (*wIter) * points[i].X();
        sumY += (*wIter) * points[i].Y();
        sumXX += (*wIter) * points[i].X() * points[i].X();
        sumXY += (*wIter) * points[i].X() * points[i].Y();
        sumW += (*wIter);
      }
      //create matrices to solve Ax = B
      double A[2][2] =
      {
        {sumXX, sumX},
        {sumX, sumW}
      };
      double B[2] =
      {
        sumXY,
        sumY
      };
      double X[2];
 
      bool nonsingular = Wm5::LinearSystem<double>().Solve2(A, B, X);
      if (nonsingular)
      {
        slope_ = X[0];
        intercept_ = X[1];
      }
      chi_squared_ = computeWeightedChiSquare(x_begin, x_end, y_begin, w_begin, slope_, intercept_);
 
      if (nonsingular)
      {
        if (compute_goodness && points.size() > 2) computeGoodness_(points, confidence_interval_P);
      }
      else
      {
        throw Exception::UnableToFit(__FILE__, __LINE__, OPENMS_PRETTY_FUNCTION,
            "UnableToFit-LinearRegression", "Could not fit a linear model to the data");
      }
    }
 
  } // namespace Math
} // namespace OpenMS