From d49e1ad5e9ec120643936d6841996efe5b271d84 Mon Sep 17 00:00:00 2001
From: Rainer Kartmann <rainer.kartmann@kit.edu>
Date: Thu, 8 Oct 2020 10:11:51 +0200
Subject: [PATCH] Redirect stats in VirtualRobot to SimoxUtility
---
VirtualRobot/CMakeLists.txt | 2 -
VirtualRobot/math/statistics/BoxPlot.cpp | 60 -------
VirtualRobot/math/statistics/BoxPlot.h | 31 +---
VirtualRobot/math/statistics/Histogram.cpp | 177 ---------------------
VirtualRobot/math/statistics/Histogram.h | 92 +----------
VirtualRobot/math/statistics/measures.cpp | 78 ++-------
6 files changed, 22 insertions(+), 418 deletions(-)
delete mode 100644 VirtualRobot/math/statistics/BoxPlot.cpp
delete mode 100644 VirtualRobot/math/statistics/Histogram.cpp
diff --git a/VirtualRobot/CMakeLists.txt b/VirtualRobot/CMakeLists.txt
index 8e32dec13..738f6bd2a 100644
--- a/VirtualRobot/CMakeLists.txt
+++ b/VirtualRobot/CMakeLists.txt
@@ -271,8 +271,6 @@ SET(SOURCES
math/Triangle.cpp
math/WeightedAverage.cpp
math/statistics/measures.cpp
- math/statistics/BoxPlot.cpp
- math/statistics/Histogram.cpp
MJCF/Document.cpp
MJCF/elements/core/AnyElement.cpp
diff --git a/VirtualRobot/math/statistics/BoxPlot.cpp b/VirtualRobot/math/statistics/BoxPlot.cpp
deleted file mode 100644
index 08b5445a5..000000000
--- a/VirtualRobot/math/statistics/BoxPlot.cpp
+++ /dev/null
@@ -1,60 +0,0 @@
-#include "BoxPlot.h"
-
-#include "measures.h"
-
-
-namespace math { namespace stat
-{
-
-
-BoxPlot::BoxPlot() = default;
-
-
-BoxPlot::BoxPlot(const std::vector<float>& values, bool isSorted, float whisk) :
- whisk(whisk)
-{
- set(values, isSorted);
-}
-
-void BoxPlot::set(const std::vector<float>& _values, bool isSorted)
-{
- const std::vector<float>& values = isSorted ? _values : sorted(_values);
-
- this->minimum = stat::min(values, true);
- this->maximum = stat::max(values, true);
-
- this->lowerQuartile = stat::lowerQuartile(values, true);
- this->median = stat::median(values, true);
- this->upperQuartile = stat::upperQuartile(values, true);
-
- const float iqr = interquartileRange(lowerQuartile, upperQuartile);
-
- this->minWhisker = lowerQuartile - whisk * iqr;
- this->maxWhisker = upperQuartile + whisk * iqr;
-
-
- // compute outliers and correct whiskers if necessary
-
- {
- auto it = values.begin();
- for (; it != values.end() && *it < minWhisker; ++it)
- {
- outliers.push_back(*it);
- }
- minWhisker = (it != values.begin()) ? *it : minimum;
- }
-
- {
- auto rit = values.rbegin();
- for (; rit != values.rend() && *rit > maxWhisker; ++rit)
- {
- outliers.push_back(*rit);
- }
- maxWhisker = (rit != values.rbegin()) ? *rit : maximum;
- }
-
-}
-
-
-
-}}
diff --git a/VirtualRobot/math/statistics/BoxPlot.h b/VirtualRobot/math/statistics/BoxPlot.h
index cd9089d6e..5c76ebe82 100644
--- a/VirtualRobot/math/statistics/BoxPlot.h
+++ b/VirtualRobot/math/statistics/BoxPlot.h
@@ -1,37 +1,10 @@
#pragma once
-#include <vector>
+#include <SimoxUtility/math/statistics/BoxPlotStats.h>
namespace math { namespace stat
{
-
- /**
- * @brief Computes and stores statistical measures found in a box plot.
- */
- class BoxPlot
- {
- public:
-
-
- BoxPlot();
- BoxPlot(const std::vector<float>& values, bool isSorted = false, float whisk = 1.5);
-
- void set(const std::vector<float>& values, bool isSorted = false);
-
- float whisk = 1.5;
-
- float minimum;
- float minWhisker;
- float lowerQuartile;
- float median;
- float upperQuartile;
- float maxWhisker;
- float maximum;
-
- std::vector<float> outliers;
-
- };
-
+ using BoxPlot = simox::math::BoxPlotStats;
}}
diff --git a/VirtualRobot/math/statistics/Histogram.cpp b/VirtualRobot/math/statistics/Histogram.cpp
deleted file mode 100644
index 3ec3cd128..000000000
--- a/VirtualRobot/math/statistics/Histogram.cpp
+++ /dev/null
@@ -1,177 +0,0 @@
-#include "Histogram.h"
-
-#include <algorithm>
-
-
-namespace VirtualRobot
-{
-
-Histogram::Histogram()
-{}
-
-Histogram::Histogram(const std::vector<float>& data, std::size_t numBins)
-{
- setMinMax(data);
- resetBins(numBins);
- insert(data);
-}
-
-Histogram::Histogram(const std::vector<float>& data, float min, float max, std::size_t numBins)
-{
- setMinMax(min, max);
- resetBins(numBins);
- insert(data);
-}
-
-void Histogram::resetBins(std::size_t numBins)
-{
- bins.assign(numBins, 0);
-}
-
-std::size_t Histogram::valueToIndex(float value) const
-{
- // normalized = 0..1
- float normalized = (value - min) / (max - min);
- std::size_t index = static_cast<std::size_t>(normalized * getNumberOfBins());
-
- // avoid out-of-bounce errors
- // (if histogram is falsely used, this could lead to peaks at the edges)
- return std::max(std::size_t(0), std::min(getNumberOfBins() - 1, index));
-}
-
-float Histogram::indexToValue(std::size_t index) const
-{
- float normalized = static_cast<float>(index) / getNumberOfBins();
- float value = normalized * (max - min) + min;
- return value;
-}
-
-void Histogram::insert(float value)
-{
- bins[valueToIndex(value)]++;
-}
-
-void Histogram::insert(const std::vector<float>& values)
-{
- for (float v : values)
- {
- insert(v);
- }
-}
-
-void Histogram::setMinMax(float min, float max)
-{
- this->min = min;
- this->max = max;
-}
-
-void Histogram::setMinMax(const std::vector<float>& data)
-{
- float min = data.front();
- float max = data.front();
-
- for (float d : data)
- {
- min = std::min(min, d);
- max = std::max(max, d);
- }
- setMinMax(min, max);
-}
-
-const std::vector<std::size_t>& Histogram::getBins() const
-{
- return bins;
-}
-
-std::size_t Histogram::getNumberOfBins() const
-{
- return bins.size();
-}
-
-float Histogram::getMin() const
-{
- return min;
-}
-
-float Histogram::getMax() const
-{
- return max;
-}
-
-std::size_t Histogram::getMinBinIndex() const
-{
- return static_cast<std::size_t>(
- std::distance(bins.begin(), std::min_element(bins.begin(), bins.end())));
-}
-
-float Histogram::getMinBinValue() const
-{
- return indexToValue(getMaxBinIndex());
-}
-
-std::size_t Histogram::getMaxBinIndex() const
-{
- return static_cast<std::size_t>(
- std::distance(bins.begin(), std::max_element(bins.begin(), bins.end())));
-}
-
-float Histogram::getMaxBinValue() const
-{
- return indexToValue(getMaxBinIndex());
-}
-
-void Histogram::applyMedianFilter(std::size_t size)
-{
- std::vector<std::size_t> newBins(bins.size());
- std::vector<std::size_t> neighborhood(2 * size + 1);
-
- for (std::size_t index = 0; index < bins.size(); index++)
- {
- // handle cases at borders
- std::size_t beginIndex = std::max(index - size, std::size_t(0));
- std::size_t endIndex = std::min(index + size + 1, bins.size());
- std::size_t num = endIndex - beginIndex; // common case: num == 2*size+1
-
- // example: index = 3, size = 2
- // => beginIndex = 3-2 = 1, endIndex = 3+2+1 = 6, num = 6-1 = 5
- // => neighborhood = [1, 6) == [1,5]
-
- neighborhood.assign(bins.begin() + beginIndex, bins.begin() + endIndex);
-
- std::sort(neighborhood.begin(), neighborhood.begin() + num);
-
- newBins[index] = neighborhood[num / 2]; // == median
- // (ignore cases where num is even, this only happens at the borders)
- }
-
- bins = newBins;
-}
-
-std::ostream& operator<<(std::ostream& os, const Histogram& histo)
-{
- os << "Histogram:\n";
- os << "min:;" << histo.min << ";max:;" << histo.max << ";\n";
-
- os << "Index:;";
- for (std::size_t index = 0; index < histo.bins.size(); index++)
- {
- os << index << ";";
- }
- os << "\n";
- os << "Values:;";
- for (std::size_t index = 0; index < histo.bins.size(); index++)
- {
- os << histo.indexToValue(index) << ";";
- }
- os << "\n";
- os << "Counts:;";
- for (std::size_t index = 0; index < histo.bins.size(); index++)
- {
- os << histo.bins[index] << ";";
- }
- os << "\n";
-
- return os;
-}
-
-}
diff --git a/VirtualRobot/math/statistics/Histogram.h b/VirtualRobot/math/statistics/Histogram.h
index 9ecdc0162..8a369ffd8 100644
--- a/VirtualRobot/math/statistics/Histogram.h
+++ b/VirtualRobot/math/statistics/Histogram.h
@@ -1,98 +1,14 @@
#pragma once
-#include <vector>
-#include <ostream>
+#include <SimoxUtility/math/statistics/Histogram.h>
namespace VirtualRobot
{
- /**
- * @brief Histogram for one-dimensional float data.
- */
- class Histogram
+ namespace math { namespace stat
{
- public:
-
- /// No initialization constructor.
- Histogram();
-
- /// Construct with the given data.
- /// Minimum and maximum are derived automatically.
- Histogram(const std::vector<float>& data, std::size_t numBins = 128);
-
- /// Construct with the given data and given limits.
- Histogram(const std::vector<float>& data, float min, float max, std::size_t numBins = 128);
-
-
- /// Set the number of bins and set them to 0.
- void resetBins(std::size_t numBins);
-
- /// Insert the given value into the histogram.
- void insert(float value);
- /// Inserts the given values into the histogram.
- void insert(const std::vector<float>& value);
-
- /// Set the limits.
- void setMinMax(float min, float max);
- /// Set the limits from the given data.
- void setMinMax(const std::vector<float>& data);
-
-
- /// Get the bins.
- const std::vector<std::size_t>& getBins() const;
-
- /// Get the number of bins.
- std::size_t getNumberOfBins() const;
-
- /// Transfer the given value to its corresponding bin index.
- std::size_t valueToIndex(float value) const;
- /// Get the value at the center of the bin with the given index.
- float indexToValue(std::size_t index) const;
-
- /// Get the minimum, i.e. the lower limit.
- float getMin() const;
- /// Get the maximum, i.e. the upper limit.
- float getMax() const;
-
- /// Return the index of the bin containing the fewest data points.
- std::size_t getMinBinIndex() const;
- /// Returns the corresponding value of the bin containing the fewest data points.
- float getMinBinValue() const;
-
- /// Return the index of the bin containing the most data points.
- std::size_t getMaxBinIndex() const;
- /// Returns the corresponding value of the bin containing the most data points.
- float getMaxBinValue() const;
-
-
- /**
- * @brief Applies a median filter to the histogram bins.
- *
- * The size specifies the number of neighours (in each direction)
- * considered.
- * That is, if k = size, 2k+1 values are considered for each point
- * (k neighbours in each direction).
- *
- * @param size the size of the neighbourhood (in each direction)
- */
- void applyMedianFilter(std::size_t size = 2);
-
-
- /// Streams a CVS-like description of the histogram containing the bin data.
- friend std::ostream& operator<<(std::ostream& os, const Histogram& histo);
-
-
- private:
-
- /// The minimum mapped value.
- float min;
- /// The maximum mapped value.
- float max;
-
- /// The bins.
- std::vector<std::size_t> bins;
-
- };
+ using Histogram = simox::math::Histogram;
+ }}
}
diff --git a/VirtualRobot/math/statistics/measures.cpp b/VirtualRobot/math/statistics/measures.cpp
index 1430b7b22..73aa17070 100644
--- a/VirtualRobot/math/statistics/measures.cpp
+++ b/VirtualRobot/math/statistics/measures.cpp
@@ -1,120 +1,74 @@
#include "measures.h"
-#include <algorithm>
-#include <cmath>
-#include <stdexcept>
+#include <SimoxUtility/math/statistics/measures.h>
namespace math { namespace stat
{
-static void checkNotEmpty(const std::vector<float>& values)
-{
- if (values.empty())
- throw std::range_error("Given values are empty.");
-}
-
void sort(std::vector<float>& values)
{
- std::sort(values.begin(), values.end());
+ simox::math::sort(values);
}
std::vector<float> sorted(const std::vector<float>& values)
{
- std::vector<float> s = values;
- std::sort(s.begin(), s.end());
- return s;
+ return simox::math::sorted(values);
}
float min(const std::vector<float>& values, bool isSorted)
{
- checkNotEmpty(values);
- return isSorted ? values.front() : *std::min_element(values.begin(), values.end());
+ return simox::math::min(values, isSorted);
}
float max(const std::vector<float>& values, bool isSorted)
{
- checkNotEmpty(values);
- return isSorted ? values.back() : *std::max_element(values.begin(), values.end());
+ return simox::math::max(values, isSorted);
}
float mean(const std::vector<float>& values)
{
- checkNotEmpty(values);
-
- float sum = 0;
- for (float v : values)
- {
- sum += v;
- }
- return sum / values.size();
+ return simox::math::mean(values);
}
float stddev(const std::vector<float>& values)
{
- return stddev(values, mean(values));
+ return simox::math::stddev(values);
}
float stddev(const std::vector<float>& values, float mean)
{
- checkNotEmpty(values);
- float sum = 0;
- for (float v : values)
- {
- float diff = v - mean;
- sum += diff * diff;
- }
- float variance = sum / (values.size() - 1);
- return std::sqrt(variance);
+ return simox::math::stddev(values, mean);
}
-float quantile(const std::vector<float>& _values, float p, bool isSorted)
+float quantile(const std::vector<float>& values, float p, bool isSorted)
{
- checkNotEmpty(_values);
- const std::vector<float>& values = isSorted ? _values : sorted(_values);
-
- float location = p < 1 ? p * values.size() : values.size() - 1;
-
- std::size_t floor = static_cast<std::size_t>(std::floor(location));
- std::size_t ceil = static_cast<std::size_t>(std::ceil(location));
-
- if (floor == ceil)
- {
- return values.at(floor);
- }
- else
- {
- float t = location - floor;
- return (1 - t) * values.at(floor) + t * values.at(ceil);
- }
+ return simox::math::quantile(values, p, isSorted);
}
float lowerQuartile(const std::vector<float>& values, bool isSorted)
{
- return quantile(values, .25, isSorted);
+ return simox::math::lower_quartile(values, isSorted);
}
float median(const std::vector<float>& values, bool isSorted)
{
- return quantile(values, .5, isSorted);
+ return simox::math::median(values, isSorted);
}
float upperQuartile(const std::vector<float>& values, bool isSorted)
{
- return quantile(values, .75, isSorted);
+ return simox::math::upper_quartile(values, isSorted);
}
-float interquartileRange(const std::vector<float>& _values, bool isSorted)
+float interquartileRange(const std::vector<float>& values, bool isSorted)
{
- checkNotEmpty(_values);
-
- const std::vector<float>& values = isSorted ? _values : sorted(_values);
- return interquartileRange(lowerQuartile(values, true), upperQuartile(values, true));
+ return simox::math::interquartile_range(values, isSorted);
}
float interquartileRange(float lowerQuartile, float upperQuartile)
{
- return upperQuartile - lowerQuartile;
+ return simox::math::interquartile_range(lowerQuartile, upperQuartile);
}
--
GitLab