Move old statistic module to statistics_lite, a new statistics module can compute percentiles
This commit is contained in:
parent
ee2f5960f4
commit
9aa494ab96
1 changed files with 116 additions and 10 deletions
|
|
@ -3,6 +3,9 @@
|
|||
#include <iostream>
|
||||
#include <cmath>
|
||||
#include <limits>
|
||||
#include <vector>
|
||||
#include <numeric>
|
||||
#include <algorithm>
|
||||
|
||||
namespace psemek::util
|
||||
{
|
||||
|
|
@ -39,7 +42,7 @@ namespace psemek::util
|
|||
}
|
||||
|
||||
template <typename T>
|
||||
struct statistics
|
||||
struct statistics_lite
|
||||
{
|
||||
void push(T const & value);
|
||||
|
||||
|
|
@ -50,7 +53,7 @@ namespace psemek::util
|
|||
T max() const { return max_; }
|
||||
|
||||
template <typename H>
|
||||
friend statistics<H> merge(statistics<H> const & s1, statistics<H> const & s2);
|
||||
friend statistics_lite<H> merge(statistics_lite<H> const & s1, statistics_lite<H> const & s2);
|
||||
|
||||
private:
|
||||
std::size_t count_ = 0;
|
||||
|
|
@ -61,7 +64,7 @@ namespace psemek::util
|
|||
};
|
||||
|
||||
template <typename T>
|
||||
void statistics<T>::push(T const & value)
|
||||
void statistics_lite<T>::push(T const & value)
|
||||
{
|
||||
++count_;
|
||||
sum_ += value;
|
||||
|
|
@ -71,18 +74,114 @@ namespace psemek::util
|
|||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::mean() const
|
||||
T statistics_lite<T>::mean() const
|
||||
{
|
||||
return sum_ / count_;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::var() const
|
||||
T statistics_lite<T>::var() const
|
||||
{
|
||||
T const m = mean();
|
||||
return std::sqrt(sum_sqr_ / count_ - m * m);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
std::ostream & operator << (std::ostream & os, statistics_lite<T> const & s)
|
||||
{
|
||||
os << "mean = " << s.mean() << ", var = " << s.var() << ", range = [" << s.min() << " .. " << s.max() << "]";
|
||||
return os;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
statistics_lite<T> merge(statistics_lite<T> const & s1, statistics_lite<T> const & s2)
|
||||
{
|
||||
statistics_lite<T> result;
|
||||
result.count_ = s1.count_ + s2.count_;
|
||||
result.sum_ = s1.sum_ + s2.sum_;
|
||||
result.sum_sqr_ = s1.sum_sqr_ + s2.sum_sqr_;
|
||||
result.min_ = std::min(s1.min_, s2.min_);
|
||||
result.max_ = std::max(s1.max_, s2.max_);
|
||||
return result;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
struct statistics
|
||||
{
|
||||
void push(T const & value);
|
||||
|
||||
std::size_t count() const { return values_.size(); }
|
||||
T mean() const;
|
||||
T var() const;
|
||||
T min() const;
|
||||
T max() const;
|
||||
|
||||
T percentile(double p) const;
|
||||
|
||||
template <typename H>
|
||||
friend statistics<H> merge(statistics<H> const & s1, statistics<H> const & s2);
|
||||
|
||||
private:
|
||||
mutable std::vector<T> values_;
|
||||
mutable bool sorted_ = true;
|
||||
};
|
||||
|
||||
template <typename T>
|
||||
void statistics<T>::push(T const & value)
|
||||
{
|
||||
values_.push_back(value);
|
||||
sorted_ = false;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::mean() const
|
||||
{
|
||||
return std::accumulate(values_.begin(), values_.end(), T{}) / count();
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::var() const
|
||||
{
|
||||
T sum{};
|
||||
T sum_sqr{};
|
||||
for (auto const & v : values_)
|
||||
{
|
||||
sum += v;
|
||||
sum_sqr += v * v;
|
||||
}
|
||||
|
||||
auto m = sum / count();
|
||||
return std::sqrt(sum_sqr / count() - m * m);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::min() const
|
||||
{
|
||||
if (values_.empty())
|
||||
return std::numeric_limits<T>::max();
|
||||
return *std::min_element(values_.begin(), values_.end());
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::max() const
|
||||
{
|
||||
if (values_.empty())
|
||||
return std::numeric_limits<T>::min();
|
||||
return *std::max_element(values_.begin(), values_.end());
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
T statistics<T>::percentile(double p) const
|
||||
{
|
||||
if (!sorted_)
|
||||
{
|
||||
std::sort(values_.begin(), values_.end());
|
||||
sorted_ = true;
|
||||
}
|
||||
|
||||
return values_[std::min<std::size_t>(values_.size() - 1, values_.size() * p)];
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
std::ostream & operator << (std::ostream & os, statistics<T> const & s)
|
||||
{
|
||||
|
|
@ -94,11 +193,18 @@ namespace psemek::util
|
|||
statistics<T> merge(statistics<T> const & s1, statistics<T> const & s2)
|
||||
{
|
||||
statistics<T> result;
|
||||
result.count_ = s1.count_ + s2.count_;
|
||||
result.sum_ = s1.sum_ + s2.sum_;
|
||||
result.sum_sqr_ = s1.sum_sqr_ + s2.sum_sqr_;
|
||||
result.min_ = std::min(s1.min_, s2.min_);
|
||||
result.max_ = std::max(s1.max_, s2.max_);
|
||||
result.values_.reserve(s1.values_.size() + s2.values_.size());
|
||||
result.sorted_ = s1.sorted_ && s2.sorted_;
|
||||
if (result.sorted_)
|
||||
{
|
||||
std::merge(s1.values_.begin(), s1.values_.end(), s2.values_.begin(), s2.values_.end(), std::back_inserter(result.values_));
|
||||
}
|
||||
else
|
||||
{
|
||||
auto it = std::back_inserter(result.values_);
|
||||
it = std::copy(s1.values_.begin(), s1.values_.end(), it);
|
||||
it = std::copy(s2.values_.begin(), s2.values_.end(), it);
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue