2998 |
14 Mar 13 |
peter |
1 |
#ifndef theplu_yat_statistics_kolmogorov_smirnov_one_sample |
2998 |
14 Mar 13 |
peter |
2 |
#define theplu_yat_statistics_kolmogorov_smirnov_one_sample |
2998 |
14 Mar 13 |
peter |
3 |
|
2998 |
14 Mar 13 |
peter |
// $Id$ |
2998 |
14 Mar 13 |
peter |
5 |
|
2998 |
14 Mar 13 |
peter |
6 |
/* |
2998 |
14 Mar 13 |
peter |
Copyright (C) 2013 Peter Johansson |
2998 |
14 Mar 13 |
peter |
8 |
|
2998 |
14 Mar 13 |
peter |
This file is part of the yat library, http://dev.thep.lu.se/yat |
2998 |
14 Mar 13 |
peter |
10 |
|
2998 |
14 Mar 13 |
peter |
The yat library is free software; you can redistribute it and/or |
2998 |
14 Mar 13 |
peter |
modify it under the terms of the GNU General Public License as |
2998 |
14 Mar 13 |
peter |
published by the Free Software Foundation; either version 3 of the |
2998 |
14 Mar 13 |
peter |
License, or (at your option) any later version. |
2998 |
14 Mar 13 |
peter |
15 |
|
2998 |
14 Mar 13 |
peter |
The yat library is distributed in the hope that it will be useful, |
2998 |
14 Mar 13 |
peter |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
2998 |
14 Mar 13 |
peter |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
2998 |
14 Mar 13 |
peter |
General Public License for more details. |
2998 |
14 Mar 13 |
peter |
20 |
|
2998 |
14 Mar 13 |
peter |
You should have received a copy of the GNU General Public License |
2998 |
14 Mar 13 |
peter |
along with yat. If not, see <http://www.gnu.org/licenses/>. |
2998 |
14 Mar 13 |
peter |
23 |
*/ |
2998 |
14 Mar 13 |
peter |
24 |
|
2998 |
14 Mar 13 |
peter |
25 |
#include <iosfwd> |
2998 |
14 Mar 13 |
peter |
26 |
|
2998 |
14 Mar 13 |
peter |
27 |
#include <map> |
2998 |
14 Mar 13 |
peter |
28 |
|
2998 |
14 Mar 13 |
peter |
29 |
namespace theplu { |
2998 |
14 Mar 13 |
peter |
30 |
namespace yat { |
2998 |
14 Mar 13 |
peter |
31 |
namespace statistics { |
2998 |
14 Mar 13 |
peter |
32 |
|
2998 |
14 Mar 13 |
peter |
33 |
/** |
2998 |
14 Mar 13 |
peter |
\brief Kolmogorov Smirnov Test for one class |
2998 |
14 Mar 13 |
peter |
35 |
|
2998 |
14 Mar 13 |
peter |
Class can be used to test if data follows standard uniform |
2998 |
14 Mar 13 |
peter |
distribution. Other distributions are not supported directly, but |
2998 |
14 Mar 13 |
peter |
it is possible to check against other distribution by tranforming |
2998 |
14 Mar 13 |
peter |
the data. If, for example, X is standard exponentially |
2998 |
14 Mar 13 |
peter |
distributed log(X) is standard uniform. Therefore, rather than |
2998 |
14 Mar 13 |
peter |
testing if X is exponential it is equivalent to test if log(X) is |
2998 |
14 Mar 13 |
peter |
standard uniform. |
2998 |
14 Mar 13 |
peter |
43 |
|
2998 |
14 Mar 13 |
peter |
\since New in yat 0.11 |
2998 |
14 Mar 13 |
peter |
45 |
*/ |
2998 |
14 Mar 13 |
peter |
46 |
class KolmogorovSmirnovOneSample |
2998 |
14 Mar 13 |
peter |
47 |
{ |
2998 |
14 Mar 13 |
peter |
48 |
public: |
2998 |
14 Mar 13 |
peter |
49 |
/** |
2998 |
14 Mar 13 |
peter |
\brief Constructor |
2998 |
14 Mar 13 |
peter |
51 |
*/ |
2998 |
14 Mar 13 |
peter |
52 |
KolmogorovSmirnovOneSample(void); |
2998 |
14 Mar 13 |
peter |
53 |
|
2998 |
14 Mar 13 |
peter |
54 |
/** |
2998 |
14 Mar 13 |
peter |
\brief add a value |
2998 |
14 Mar 13 |
peter |
56 |
|
2998 |
14 Mar 13 |
peter |
\a value should be a value in range [0, 1]. |
2998 |
14 Mar 13 |
peter |
58 |
*/ |
2998 |
14 Mar 13 |
peter |
59 |
void add(double value, double weight=1.0); |
2998 |
14 Mar 13 |
peter |
60 |
|
2998 |
14 Mar 13 |
peter |
61 |
/** |
2998 |
14 Mar 13 |
peter |
\brief Large-Sample Approximation |
2998 |
14 Mar 13 |
peter |
63 |
|
2998 |
14 Mar 13 |
peter |
This analytical approximation of p-value can be used when all |
2998 |
14 Mar 13 |
peter |
weight equal unity and sample size \a n is large. The p-value |
2998 |
14 Mar 13 |
peter |
is calcuated as \f$ P = \displaystyle - 2 \sum_{k=1}^{\infty} |
3001 |
19 Mar 13 |
peter |
(-1)^ke^{-2k^2s^2n}\f$, where s is the score and n is the sum |
3001 |
19 Mar 13 |
peter |
of weights. |
2998 |
14 Mar 13 |
peter |
69 |
*/ |
2998 |
14 Mar 13 |
peter |
70 |
double p_value(void) const; |
2998 |
14 Mar 13 |
peter |
71 |
|
2998 |
14 Mar 13 |
peter |
72 |
/** |
2998 |
14 Mar 13 |
peter |
\brief Remove a data point |
2998 |
14 Mar 13 |
peter |
74 |
|
2998 |
14 Mar 13 |
peter |
\throw utility::runtime_error if no data point exist with \a |
2998 |
14 Mar 13 |
peter |
value and \a weight. |
2998 |
14 Mar 13 |
peter |
77 |
*/ |
2998 |
14 Mar 13 |
peter |
78 |
void remove(double value, double weight=1.0); |
2998 |
14 Mar 13 |
peter |
79 |
|
2998 |
14 Mar 13 |
peter |
80 |
/** |
2998 |
14 Mar 13 |
peter |
\brief resets everything to zero |
2998 |
14 Mar 13 |
peter |
82 |
*/ |
2998 |
14 Mar 13 |
peter |
83 |
void reset(void); |
2998 |
14 Mar 13 |
peter |
84 |
|
2998 |
14 Mar 13 |
peter |
85 |
/** |
2998 |
14 Mar 13 |
peter |
\brief Kolmogorov Smirnov statistic |
2998 |
14 Mar 13 |
peter |
87 |
|
2998 |
14 Mar 13 |
peter |
\f$ sup_x | F(x) - x | \f$ where is the empirical cumulative |
2998 |
14 Mar 13 |
peter |
distribution \f$ F(x) = \frac{\sum_{i:x_i\le x}w_i}{ \sum w_i} |
2998 |
14 Mar 13 |
peter |
\f$ |
2998 |
14 Mar 13 |
peter |
91 |
*/ |
2998 |
14 Mar 13 |
peter |
92 |
double score(void) const; |
2998 |
14 Mar 13 |
peter |
93 |
|
2998 |
14 Mar 13 |
peter |
94 |
/** |
2998 |
14 Mar 13 |
peter |
Same as score() but keeping the sign, in other words, |
2998 |
14 Mar 13 |
peter |
abs(signed_score())==score() |
2998 |
14 Mar 13 |
peter |
97 |
|
2998 |
14 Mar 13 |
peter |
A positive score implies that values \em on \em average are |
2998 |
14 Mar 13 |
peter |
smaller 0.5. |
2998 |
14 Mar 13 |
peter |
100 |
*/ |
2998 |
14 Mar 13 |
peter |
101 |
double signed_score(void) const; |
2998 |
14 Mar 13 |
peter |
102 |
|
2998 |
14 Mar 13 |
peter |
103 |
private: |
2998 |
14 Mar 13 |
peter |
104 |
mutable double score_; |
2998 |
14 Mar 13 |
peter |
105 |
mutable bool cached_; |
2998 |
14 Mar 13 |
peter |
106 |
std::multimap<double, double> data_; |
2998 |
14 Mar 13 |
peter |
107 |
double sum_w_; |
2998 |
14 Mar 13 |
peter |
108 |
|
2998 |
14 Mar 13 |
peter |
109 |
friend std::ostream& operator<<(std::ostream&, |
2998 |
14 Mar 13 |
peter |
110 |
const KolmogorovSmirnovOneSample&); |
2998 |
14 Mar 13 |
peter |
111 |
|
2998 |
14 Mar 13 |
peter |
// using compiler generated copy and assignment |
2998 |
14 Mar 13 |
peter |
//KolmogorovSmirnovOneSample(const KolmogorovSmirnovOneSample&); |
2998 |
14 Mar 13 |
peter |
//KolmogorovSmirnovOneSample& operator=(const KolmogorovSmirnovOneSample&); |
2998 |
14 Mar 13 |
peter |
115 |
}; |
2998 |
14 Mar 13 |
peter |
116 |
|
2998 |
14 Mar 13 |
peter |
117 |
}}} // of namespace theplu yat statistics |
2998 |
14 Mar 13 |
peter |
118 |
|
2998 |
14 Mar 13 |
peter |
119 |
#endif |