Changeset 718 for trunk/yat/statistics/ROC.h
 Timestamp:
 Dec 26, 2006, 10:56:26 AM (16 years ago)
 File:

 1 edited
Legend:
 Unmodified
 Added
 Removed

trunk/yat/statistics/ROC.h
r703 r718 59 59 virtual ~ROC(void); 60 60 61 /// 62 /// minimum_size is the threshold for when a normal 63 /// approximation is used for the pvalue calculation. 64 /// 65 /// @return reference to minimum_size 66 /// 67 u_int& minimum_size(void); 68 69 /// 70 /// @return number of samples 71 /// 72 size_t n(void) const; 73 74 /// 75 /// @return number of positive samples (Target.binary()==true) 76 /// 77 size_t n_pos(void) const; 78 79 /// 80 ///Calculates the pvalue, i.e. the probability of observing an 81 ///area equally or larger if the null hypothesis is true. If P is 82 ///near zero, this casts doubt on this hypothesis. The null 83 ///hypothesis is that the values from the 2 classes are generated 84 ///from 2 identical distributions. The alternative is that the 85 ///median of the first distribution is shifted from the median of 86 ///the second distribution by a nonzero amount. If the smallest 87 ///group size is larger than minimum_size (default = 10), then P 88 ///is calculated using a normal approximation. @return the 89 ///onesided pvalue( if absolute true is used this is equivalent 90 ///to the twosided pvalue.) 91 /// 92 double p_value(void) const; 93 61 94 /// Function taking \a value, \a target (+1 or 1) and vector 62 95 /// defining what samples to use. The score is equivalent to … … 83 116 double score(const classifier::Target& target, 84 117 const classifier::DataLookupWeighted1D& value); 85 86 118 87 119 /** … … 100 132 const utility::vector& value, 101 133 const utility::vector& weight); 102 103 104 ///105 ///Calculates the pvalue, i.e. the probability of observing an106 ///area equally or larger if the null hypothesis is true. If P is107 ///near zero, this casts doubt on this hypothesis. The null108 ///hypothesis is that the values from the 2 classes are generated109 ///from 2 identical distributions. The alternative is that the110 ///median of the first distribution is shifted from the median of111 ///the second distribution by a nonzero amount. If the smallest112 ///group size is larger than minimum_size (default = 10), then P113 ///is calculated using a normal approximation. @return the114 ///onesided pvalue( if absolute true is used this is equivalent115 ///to the twosided pvalue.)116 ///117 double p_value(void) const;118 119 ///120 /// minimum_size is the threshold for when a normal121 /// approximation is used for the pvalue calculation.122 ///123 /// @return reference to minimum_size124 ///125 inline u_int& minimum_size(void){ return minimum_size_; }126 134 127 135 /// … … 132 140 /// 133 141 bool target(const size_t i) const; 134 135 ///136 /// @return number of samples137 ///138 inline size_t n(void) const { return vec_pair_.size(); }139 140 ///141 /// @return number of positive samples (Target.binary()==true)142 ///143 inline size_t n_pos(void) const { return nof_pos_; }144 142 145 143 private:
Note: See TracChangeset
for help on using the changeset viewer.