2 ///////////////////////////////////////////////////////////////////////////////
5 // Copyright 2006 Daniel Egloff, Olivier Gygi. Distributed under the Boost
6 // Software License, Version 1.0. (See accompanying file
7 // LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
9 #ifndef BOOST_ACCUMULATORS_STATISTICS_DENSITY_HPP_DE_01_01_2006
10 #define BOOST_ACCUMULATORS_STATISTICS_DENSITY_HPP_DE_01_01_2006
15 #include <boost/range.hpp>
16 #include <boost/parameter/keyword.hpp>
17 #include <boost/mpl/placeholders.hpp>
18 #include <boost/accumulators/accumulators_fwd.hpp>
19 #include <boost/accumulators/framework/accumulator_base.hpp>
20 #include <boost/accumulators/framework/extractor.hpp>
21 #include <boost/accumulators/numeric/functional.hpp>
22 #include <boost/accumulators/framework/parameters/sample.hpp>
23 #include <boost/accumulators/framework/depends_on.hpp>
24 #include <boost/accumulators/statistics_fwd.hpp>
25 #include <boost/accumulators/statistics/count.hpp>
26 #include <boost/accumulators/statistics/max.hpp>
27 #include <boost/accumulators/statistics/min.hpp>
29 namespace boost
{ namespace accumulators
32 ///////////////////////////////////////////////////////////////////////////////
33 // cache_size and num_bins named parameters
35 BOOST_PARAMETER_NESTED_KEYWORD(tag
, density_cache_size
, cache_size
)
36 BOOST_PARAMETER_NESTED_KEYWORD(tag
, density_num_bins
, num_bins
)
38 BOOST_ACCUMULATORS_IGNORE_GLOBAL(density_cache_size
)
39 BOOST_ACCUMULATORS_IGNORE_GLOBAL(density_num_bins
)
43 ///////////////////////////////////////////////////////////////////////////////
47 @brief Histogram density estimator
49 The histogram density estimator returns a histogram of the sample distribution. The positions and sizes of the bins
50 are determined using a specifiable number of cached samples (cache_size). The range between the minimum and the
51 maximum of the cached samples is subdivided into a specifiable number of bins (num_bins) of same size. Additionally,
52 an under- and an overflow bin is added to capture future under- and overflow samples. Once the bins are determined,
53 the cached samples and all subsequent samples are added to the correct bins. At the end, a range of std::pair is
54 return, where each pair contains the position of the bin (lower bound) and the samples count (normalized with the
55 total number of samples).
57 @param density_cache_size Number of first samples used to determine min and max.
58 @param density_num_bins Number of bins (two additional bins collect under- and overflow samples).
60 template<typename Sample
>
64 typedef typename
numeric::functional::fdiv
<Sample
, std::size_t>::result_type float_type
;
65 typedef std::vector
<std::pair
<float_type
, float_type
> > histogram_type
;
66 typedef std::vector
<float_type
> array_type
;
67 // for boost::result_of
68 typedef iterator_range
<typename
histogram_type::iterator
> result_type
;
70 template<typename Args
>
71 density_impl(Args
const &args
)
72 : cache_size(args
[density_cache_size
])
74 , num_bins(args
[density_num_bins
])
75 , samples_in_bin(num_bins
+ 2, 0.)
76 , bin_positions(num_bins
+ 2)
80 numeric::fdiv(args
[sample
| Sample()],(std::size_t)1)
81 , numeric::fdiv(args
[sample
| Sample()],(std::size_t)1)
88 template<typename Args
>
89 void operator ()(Args
const &args
)
91 this->is_dirty
= true;
93 std::size_t cnt
= count(args
);
95 // Fill up cache with cache_size first samples
96 if (cnt
<= this->cache_size
)
98 this->cache
[cnt
- 1] = args
[sample
];
101 // Once cache_size samples have been accumulated, create num_bins bins of same size between
102 // the minimum and maximum of the cached samples as well as under and overflow bins.
103 // Store their lower bounds (bin_positions) and fill the bins with the cached samples (samples_in_bin).
104 if (cnt
== this->cache_size
)
106 float_type minimum
= numeric::fdiv((min
)(args
), (std::size_t)1);
107 float_type maximum
= numeric::fdiv((max
)(args
), (std::size_t)1);
108 float_type bin_size
= numeric::fdiv(maximum
- minimum
, this->num_bins
);
110 // determine bin positions (their lower bounds)
111 for (std::size_t i
= 0; i
< this->num_bins
+ 2; ++i
)
113 this->bin_positions
[i
] = minimum
+ (i
- 1.) * bin_size
;
116 for (typename
array_type::const_iterator iter
= this->cache
.begin(); iter
!= this->cache
.end(); ++iter
)
118 if (*iter
< this->bin_positions
[1])
120 ++(this->samples_in_bin
[0]);
122 else if (*iter
>= this->bin_positions
[this->num_bins
+ 1])
124 ++(this->samples_in_bin
[this->num_bins
+ 1]);
128 typename
array_type::iterator it
= std::upper_bound(
129 this->bin_positions
.begin()
130 , this->bin_positions
.end()
134 std::size_t d
= std::distance(this->bin_positions
.begin(), it
);
135 ++(this->samples_in_bin
[d
- 1]);
139 // Add each subsequent sample to the correct bin
140 else if (cnt
> this->cache_size
)
142 if (args
[sample
] < this->bin_positions
[1])
144 ++(this->samples_in_bin
[0]);
146 else if (args
[sample
] >= this->bin_positions
[this->num_bins
+ 1])
148 ++(this->samples_in_bin
[this->num_bins
+ 1]);
152 typename
array_type::iterator it
= std::upper_bound(
153 this->bin_positions
.begin()
154 , this->bin_positions
.end()
158 std::size_t d
= std::distance(this->bin_positions
.begin(), it
);
159 ++(this->samples_in_bin
[d
- 1]);
165 @pre The number of samples must meet or exceed the cache size
167 template<typename Args
>
168 result_type
result(Args
const &args
) const
172 this->is_dirty
= false;
174 // creates a vector of std::pair where each pair i holds
175 // the values bin_positions[i] (x-axis of histogram) and
176 // samples_in_bin[i] / cnt (y-axis of histogram).
178 for (std::size_t i
= 0; i
< this->num_bins
+ 2; ++i
)
180 this->histogram
[i
] = std::make_pair(this->bin_positions
[i
], numeric::fdiv(this->samples_in_bin
[i
], count(args
)));
183 // returns a range of pairs
184 return make_iterator_range(this->histogram
);
188 std::size_t cache_size
; // number of cached samples
189 array_type cache
; // cache to store the first cache_size samples
190 std::size_t num_bins
; // number of bins
191 array_type samples_in_bin
; // number of samples in each bin
192 array_type bin_positions
; // lower bounds of bins
193 mutable histogram_type histogram
; // histogram
194 mutable bool is_dirty
;
199 ///////////////////////////////////////////////////////////////////////////////
205 : depends_on
<count
, min
, max
>
211 typedef accumulators::impl::density_impl
<mpl::_1
> impl
;
213 #ifdef BOOST_ACCUMULATORS_DOXYGEN_INVOKED
214 /// tag::density::cache_size named parameter
215 /// tag::density::num_bins named parameter
216 static boost::parameter::keyword
<density_cache_size
> const cache_size
;
217 static boost::parameter::keyword
<density_num_bins
> const num_bins
;
222 ///////////////////////////////////////////////////////////////////////////////
227 extractor
<tag::density
> const density
= {};
229 BOOST_ACCUMULATORS_IGNORE_GLOBAL(density
)
232 using extract::density
;
234 // So that density can be automatically substituted
235 // with weighted_density when the weight parameter is non-void.
237 struct as_weighted_feature
<tag::density
>
239 typedef tag::weighted_density type
;
243 struct feature_of
<tag::weighted_density
>
244 : feature_of
<tag::density
>
248 }} // namespace boost::accumulators