Changeset 1643 for trunk/yat/regression/GSLInterpolation.h
- Timestamp:
- Dec 13, 2008, 1:23:39 AM (14 years ago)
- File:
-
- 1 copied
Legend:
- Unmodified
- Added
- Removed
-
trunk/yat/regression/GSLInterpolation.h
r1637 r1643 1 #ifndef _theplu_yat_regression_ linear_2 #define _theplu_yat_regression_ linear_1 #ifndef _theplu_yat_regression_gslinterpolation_ 2 #define _theplu_yat_regression_gslinterpolation_ 3 3 4 4 // $Id$ 5 5 6 6 /* 7 Copyright (C) 2004, 2005, 2006, 2007 Jari Häkkinen, Peter Johansson 8 Copyright (C) 2008 Peter Johansson 7 Copyright (C) 2008 Jari Häkkinen 9 8 10 9 This file is part of the yat library, http://dev.thep.lu.se/yat … … 24 23 */ 25 24 26 #include "OneDimensional.h" 27 28 #include <cmath> 25 #include <gsl/gsl_interp.h> 29 26 30 27 namespace theplu { 31 28 namespace yat { 32 33 class VectorBase;34 29 namespace utility { 30 class VectorConstView; 31 } 35 32 namespace regression { 36 33 37 34 /** 38 @brief linear regression. 39 40 Data are modeled as \f$ y_i = \alpha + \beta (x_i-m_x) + 41 \epsilon_i \f$. 35 @brief Documentation please. 42 36 */ 43 class Linear : public OneDimensional 44 { 45 46 public: 47 /// 48 /// @brief The default constructor 49 /// 50 Linear(void); 37 class GSLInterpolation 38 { 51 39 52 /// 53 /// @brief The destructor 54 /// 55 virtual ~Linear(void); 56 57 /** 58 The parameter \f$ \alpha \f$ is estimated as \f$ 59 \frac{1}{n}\sum y_i \f$ 60 61 @return the parameter \f$ \alpha \f$ 62 */ 63 double alpha(void) const; 40 public: 64 41 65 42 /** 66 The standard deviation is estimated as \f$ \sqrt{\frac{s^2}{n}} 67 \f$ where \f$ s^2 = \frac{\sum \epsilon^2}{n-2} \f$ 68 69 @return standard deviation of parameter \f$ \alpha \f$ 43 @brief Documentation please. 70 44 */ 71 double alpha_var(void) const; 45 double evaluate(const double x) const; 46 47 protected: 48 /** 49 @brief The default constructor 50 */ 51 GSLInterpolation(const gsl_interp_type*, const utility::VectorConstView& x, 52 const utility::VectorConstView& y); 72 53 73 54 /** 74 The parameter \f$ \beta \f$ is estimated as \f$ 75 \frac{\textrm{Cov}(x,y)}{\textrm{Var}(x)} \f$ 76 77 @return the parameter \f$ \beta \f$ 55 @brief The destructor 78 56 */ 79 double beta(void) const;57 virtual ~GSLInterpolation(void); 80 58 59 private: 81 60 /** 82 The standard deviation is estimated as \f$ \frac{s^2}{\sum 83 (x-m_x)^2} \f$ where \f$ s^2 = \frac{\sum \epsilon^2}{n-2} \f$ 61 Copy Constructor. (not implemented) 62 */ 63 GSLInterpolation(const GSLInterpolation&); 84 64 85 @return standard deviation of parameter \f$ \beta \f$ 86 */ 87 double beta_var(void) const; 88 89 /** 90 Model is fitted by minimizing \f$ \sum{(y_i - \alpha - \beta 91 (x-m_x))^2} \f$. By construction \f$ \alpha \f$ and \f$ \beta \f$ 92 are independent. 93 */ 94 void fit(const utility::VectorBase& x, const utility::VectorBase& y) ; 95 96 /// 97 /// @return \f$ \alpha + \beta x \f$ 98 /// 99 double predict(const double x) const; 100 101 /** 102 \f$ \frac{\sum \epsilon_i^2}{N-2} \f$ 103 104 @return variance of residuals 105 */ 106 double s2(void) const; 107 108 /** 109 The error of the model is estimated as \f$ 110 \textrm{alpha\_err}^2+\textrm{beta\_err}^2*(x-m_x)*(x-m_x)\f$ 111 112 @return estimated error of model in @a x 113 */ 114 double standard_error2(const double x) const; 115 116 117 private: 118 /// 119 /// Copy Constructor. (not implemented) 120 /// 121 Linear(const Linear&); 122 123 double alpha_; 124 double alpha_var_; 125 double beta_; 126 double beta_var_; 65 gsl_interp_accel* accelerator_; 66 gsl_interp* interpolator_; 67 double* x_; 68 double* y_; 127 69 }; 128 70
Note: See TracChangeset
for help on using the changeset viewer.