linear_regression.hpp

Go to the documentation of this file.

#ifndef MLPACK_METHODS_LINEAR_REGRESSION_LINEAR_REGRESSION_HPP
#define MLPACK_METHODS_LINEAR_REGRESSION_LINEAR_REGRESSION_HPP

#include <mlpack/prereqs.hpp>

namespace mlpack {
namespace regression  {

class LinearRegression
{
 public:
  LinearRegression(const arma::mat& predictors,
                   const arma::rowvec& responses,
                   const double lambda = 0,
                   const bool intercept = true);

  LinearRegression(const arma::mat& predictors,
                   const arma::rowvec& responses,
                   const arma::rowvec& weights,
                   const double lambda = 0,
                   const bool intercept = true);

  LinearRegression() : lambda(0.0), intercept(true) { }

  double Train(const arma::mat& predictors,
               const arma::rowvec& responses,
               const bool intercept = true);

  double Train(const arma::mat& predictors,
               const arma::rowvec& responses,
               const arma::rowvec& weights,
               const bool intercept = true);

  void Predict(const arma::mat& points, arma::rowvec& predictions) const;

  double ComputeError(const arma::mat& points,
                      const arma::rowvec& responses) const;

  const arma::vec& Parameters() const { return parameters; }
  arma::vec& Parameters() { return parameters; }

  double Lambda() const { return lambda; }
  double& Lambda() { return lambda; }

  bool Intercept() const { return intercept; }

  template<typename Archive>
  void serialize(Archive& ar, const uint32_t /* version */)
  {
    ar(CEREAL_NVP(parameters));
    ar(CEREAL_NVP(lambda));
    ar(CEREAL_NVP(intercept));
  }

 private:
  arma::vec parameters;

  double lambda;

  bool intercept;
};

} // namespace regression
} // namespace mlpack

#endif // MLPACK_METHODS_LINEAR_REGRESSION_HPP