Add QgsStatisticalSummary class for calculating stats from a list

of values.

python/core/core.sip

@@ -103,6 +103,7 @@
 %Include qgssnapper.sip
 %Include qgssnappingutils.sip
 %Include qgsspatialindex.sip
+%Include qgsstatisticalsummary.sip
 %Include qgstolerance.sip
 %Include qgsvectordataprovider.sip
 %Include qgsvectorfilewriter.sip

python/core/qgsstatisticalsummary.sip

@@ -0,0 +1,135 @@
+/** \ingroup core
+ * \class QgsStatisticalSummary
+ * \brief Calculator for summary statistics for a list of doubles.
+ *
+ * Statistics are calculated by calling @link calculate @endlink and passing a list of doubles. The
+ * individual statistics can then be retrieved using the associated methods. Note that not all statistics
+ * are calculated by default. Statistics which require slower computations are only calculated by
+ * specifying the statistic in the constructor or via @link setStatistics @endlink.
+ *
+ * \note Added in version 2.9
+ */
+
+class QgsStatisticalSummary
+{
+%TypeHeaderCode
+#include <qgsstatisticalsummary.h>
+%End
+
+  public:
+
+    //! Enumeration of flags that specify statistics to be calculated
+    enum Statistic
+    {
+        Count ,  //!< Count
+        Sum,  //!< Sum of values
+        Mean,  //!< Mean of values
+        Median, //!< Median of values
+        StDev, //!< Standard deviation of values
+        StDevSample, //!< Sample standard deviation of values
+        Min,  //!< Min of values
+        Max,  //!< Max of values
+        Range, //!< Range of values (max - min)
+        Minority, //!< Minority of values
+        Majority, //!< Majority of values
+        Variety, //!< Variety (count of distinct) values
+        All
+    };
+    
+    typedef QFlags<QgsStatisticalSummary::Statistic> Statistics;
+
+    /** Constructor for QgsStatisticalSummary
+     * @param stats flags for statistics to calculate
+     */
+    QgsStatisticalSummary( QgsStatisticalSummary::Statistics stats = QgsStatisticalSummary::Statistics( 0 ) );
+
+    virtual ~QgsStatisticalSummary();
+
+    /** Returns flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg sum, min and max).
+     * @see setStatistics
+     */
+    QgsStatisticalSummary::Statistics statistics() const;
+
+    /** Sets flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg sum, min and max).
+     * @param stats flags for statistics to calculate
+     * @see statistics
+     */
+    void setStatistics( QgsStatisticalSummary::Statistics stats );
+
+    /** Resets the calculated values
+     */
+    void reset();
+
+    /** Calculates summary statistics for a list of values
+     * @param values list of doubles
+     */
+    void calculate( const QList<double>& values );
+
+    /** Returns calculated count of values
+     */
+    int count() const;
+
+    /** Returns calculated sum of values
+     */
+    double sum() const;
+
+    /** Returns calculated mean of values
+     */
+    double mean() const;
+
+    /** Returns calculated median of values. This is only calculated if Statistic::Median has
+     * been specified in the constructor or via setStatistics.
+     */
+    double median() const;
+
+    /** Returns calculated minimum from values.
+     */
+    double min() const;
+
+    /** Returns calculated maximum from values.
+     */
+    double max() const;
+
+    /** Returns calculated range (difference between maximum and minimum values).
+     */
+    double range() const;
+
+    /** Returns population standard deviation. This is only calculated if Statistic::StDev has
+     * been specified in the constructor or via setStatistics.
+     * @see sampleStDev
+     */
+    double stDev() const;
+
+    /** Returns sample standard deviation. This is only calculated if Statistic::StDev has
+     * been specified in the constructor or via setStatistics.
+     * @see stDev
+     */
+    double sampleStDev() const;
+
+    /** Returns variety of values. The variety is the count of unique values from the list.
+     * This is only calculated if Statistic::Variety has been specified in the constructor
+     * or via setStatistics.
+     */
+    int variety() const;
+
+    /** Returns minority of values. The minority is the value with least occurances in the list
+     * This is only calculated if Statistic::Minority has been specified in the constructor
+     * or via setStatistics.
+     * @see majority
+     */
+    double minority() const;
+
+    /** Returns majority of values. The majority is the value with most occurances in the list
+     * This is only calculated if Statistic::Majority has been specified in the constructor
+     * or via setStatistics.
+     * @see minority
+     */
+    double majority() const;
+
+};
+
+QFlags<QgsStatisticalSummary::Statistic> operator|(QgsStatisticalSummary::Statistic f1, QFlags<QgsStatisticalSummary::Statistic> f2);
+
+

src/core/CMakeLists.txt

@@ -168,6 +168,7 @@ SET(QGIS_CORE_SRCS
   qgssnapper.cpp
   qgssnappingutils.cpp
   qgsspatialindex.cpp
+  qgsstatisticalsummary.cpp
   qgstransaction.cpp
   qgstolerance.cpp
   qgsvectordataprovider.cpp
@@ -554,6 +555,7 @@ SET(QGIS_CORE_HDRS
   qgssnapper.h
   qgssnappingutils.h
   qgsspatialindex.h
+  qgsstatisticalsummary.h
   qgstolerance.h
   qgstransaction.h
   qgsvectordataprovider.h

src/core/qgsstatisticalsummary.cpp

@@ -0,0 +1,109 @@
+/***************************************************************************
+  qgsstatisticalsummary.cpp
+  --------------------------------------
+  Date                 : May 2015
+  Copyright            : (C) 2015 by Nyall Dawson
+  Email                : nyall dot dawson at gmail dot com
+ ***************************************************************************
+ *                                                                         *
+ *   This program is free software; you can redistribute it and/or modify  *
+ *   it under the terms of the GNU General Public License as published by  *
+ *   the Free Software Foundation; either version 2 of the License, or     *
+ *   (at your option) any later version.                                   *
+ *                                                                         *
+ ***************************************************************************/
+
+#include "qgsstatisticalsummary.h"
+#include <limits>
+#include <qmath.h>
+
+
+QgsStatisticalSummary::QgsStatisticalSummary( Statistics stats )
+    : mStatistics( stats )
+{
+  reset();
+}
+
+QgsStatisticalSummary::~QgsStatisticalSummary()
+{
+
+}
+
+void QgsStatisticalSummary::reset()
+{
+  mCount = 0;
+  mSum = 0;
+  mMean = 0;
+  mMedian = 0;
+  mMin = std::numeric_limits<double>::max();
+  mMax = std::numeric_limits<double>::min();
+  mStdev = 0;
+  mSampleStdev = 0;
+  mMinority = 0;
+  mMajority = 0;
+  mValueCount.clear();
+}
+
+void QgsStatisticalSummary::calculate( const QList<double> &values )
+{
+  reset();
+
+  foreach ( double value, values )
+  {
+    mCount++;
+    mSum += value;
+    mMin = qMin( mMin, value );
+    mMax = qMax( mMax, value );
+
+    if ( mStatistics & QgsStatisticalSummary::Majority || mStatistics & QgsStatisticalSummary::Minority || mStatistics & QgsStatisticalSummary::Variety )
+      mValueCount.insert( value, mValueCount.value( value, 0 ) + 1 );
+  }
+
+  if ( mCount == 0 )
+    return;
+
+  mMean = mSum / mCount;
+
+  if ( mStatistics & QgsStatisticalSummary::StDev )
+  {
+    double sumSquared = 0;
+    foreach ( double value, values )
+    {
+      double diff = value - mMean;
+      sumSquared += diff * diff;
+    }
+    mStdev = qPow( sumSquared / values.count(), 0.5 );
+    mSampleStdev = qPow( sumSquared / ( values.count() - 1 ), 0.5 );
+  }
+
+  if ( mStatistics & QgsStatisticalSummary::Median )
+  {
+    QList<double> sorted = values;
+    qSort( sorted.begin(), sorted.end() );
+    bool even = ( mCount % 2 ) < 1;
+    if ( even )
+    {
+      mMedian = ( sorted[mCount / 2 - 1] + sorted[mCount / 2] ) / 2.0;
+    }
+    else //odd
+    {
+      mMedian = sorted[( mCount + 1 ) / 2 - 1];
+    }
+  }
+
+  if ( mStatistics & QgsStatisticalSummary::Minority || mStatistics & QgsStatisticalSummary::Majority )
+  {
+    QList<int> valueCounts = mValueCount.values();
+    qSort( valueCounts.begin(), valueCounts.end() );
+    if ( mStatistics & QgsStatisticalSummary::Minority )
+    {
+      mMinority = mValueCount.key( valueCounts.first() );
+    }
+    if ( mStatistics & QgsStatisticalSummary::Majority )
+    {
+      mMajority = mValueCount.key( valueCounts.last() );
+    }
+  }
+
+}
+

src/core/qgsstatisticalsummary.h

@@ -0,0 +1,165 @@
+/***************************************************************************
+  qgsstatisticalsummary.h
+  --------------------------------------
+  Date                 : May 2015
+  Copyright            : (C) 2015 by Nyall Dawson
+  Email                : nyall dot dawson at gmail dot com
+ ***************************************************************************
+ *                                                                         *
+ *   This program is free software; you can redistribute it and/or modify  *
+ *   it under the terms of the GNU General Public License as published by  *
+ *   the Free Software Foundation; either version 2 of the License, or     *
+ *   (at your option) any later version.                                   *
+ *                                                                         *
+ ***************************************************************************/
+
+#ifndef QGSSTATISTICALSUMMARY_H
+#define QGSSTATISTICALSUMMARY_H
+
+#include <QMap>
+
+/** \ingroup core
+ * \class QgsStatisticalSummary
+ * \brief Calculator for summary statistics for a list of doubles.
+ *
+ * Statistics are calculated by calling @link calculate @endlink and passing a list of doubles. The
+ * individual statistics can then be retrieved using the associated methods. Note that not all statistics
+ * are calculated by default. Statistics which require slower computations are only calculated by
+ * specifying the statistic in the constructor or via @link setStatistics @endlink.
+ *
+ * \note Added in version 2.9
+ */
+
+class CORE_EXPORT QgsStatisticalSummary
+{
+  public:
+
+    //! Enumeration of flags that specify statistics to be calculated
+    enum Statistic
+    {
+      Count       = 1,  //!< Count
+      Sum    = 2,  //!< Sum of values
+      Mean = 4,  //!< Mean of values
+      Median = 8, //!< Median of values
+      StDev = 16, //!< Standard deviation of values
+      StDevSample = 32, //!< Sample standard deviation of values
+      Min = 64,  //!< Min of values
+      Max = 128,  //!< Max of values
+      Range = 256, //!< Range of values (max - min)
+      Minority = 512, //!< Minority of values
+      Majority = 1024, //!< Majority of values
+      Variety = 2048, //!< Variety (count of distinct) values
+      All = Count | Sum | Mean | Median | StDev | Max | Min | Range | Minority | Majority | Variety
+    };
+    Q_DECLARE_FLAGS( Statistics, Statistic )
+
+    /** Constructor for QgsStatisticalSummary
+     * @param stats flags for statistics to calculate
+     */
+    QgsStatisticalSummary( Statistics stats = Statistics( 0 ) );
+
+    virtual ~QgsStatisticalSummary();
+
+    /** Returns flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg sum, min and max).
+     * @see setStatistics
+     */
+    Statistics statistics() const { return mStatistics; }
+
+    /** Sets flags which specify which statistics will be calculated. Some statistics
+     * are always calculated (eg sum, min and max).
+     * @param stats flags for statistics to calculate
+     * @see statistics
+     */
+    void setStatistics( Statistics stats ) { mStatistics = stats; }
+
+    /** Resets the calculated values
+     */
+    void reset();
+
+    /** Calculates summary statistics for a list of values
+     * @param values list of doubles
+     */
+    void calculate( const QList<double>& values );
+
+    /** Returns calculated count of values
+     */
+    int count() const { return mCount; }
+
+    /** Returns calculated sum of values
+     */
+    double sum() const { return mSum; }
+
+    /** Returns calculated mean of values
+     */
+    double mean() const { return mMean; }
+
+    /** Returns calculated median of values. This is only calculated if Statistic::Median has
+     * been specified in the constructor or via setStatistics.
+     */
+    double median() const { return mMedian; }
+
+    /** Returns calculated minimum from values.
+     */
+    double min() const { return mMin; }
+
+    /** Returns calculated maximum from values.
+     */
+    double max() const { return mMax; }
+
+    /** Returns calculated range (difference between maximum and minimum values).
+     */
+    double range() const { return mMax - mMin; }
+
+    /** Returns population standard deviation. This is only calculated if Statistic::StDev has
+     * been specified in the constructor or via setStatistics.
+     * @see sampleStDev
+     */
+    double stDev() const { return mStdev; }
+
+    /** Returns sample standard deviation. This is only calculated if Statistic::StDev has
+     * been specified in the constructor or via setStatistics.
+     * @see stDev
+     */
+    double sampleStDev() const { return mSampleStdev; }
+
+    /** Returns variety of values. The variety is the count of unique values from the list.
+     * This is only calculated if Statistic::Variety has been specified in the constructor
+     * or via setStatistics.
+     */
+    int variety() const { return mValueCount.count(); }
+
+    /** Returns minority of values. The minority is the value with least occurances in the list
+     * This is only calculated if Statistic::Minority has been specified in the constructor
+     * or via setStatistics.
+     * @see majority
+     */
+    double minority() const { return mMinority; }
+
+    /** Returns majority of values. The majority is the value with most occurances in the list
+     * This is only calculated if Statistic::Majority has been specified in the constructor
+     * or via setStatistics.
+     * @see minority
+     */
+    double majority() const { return mMajority; }
+
+  private:
+
+    Statistics mStatistics;
+
+    int mCount;
+    double mSum;
+    double mMean;
+    double mMedian;
+    double mMin;
+    double mMax;
+    double mStdev;
+    double mSampleStdev;
+    double mMinority;
+    double mMajority;
+    QMap< double, int > mValueCount;
+};
+
+Q_DECLARE_OPERATORS_FOR_FLAGS( QgsStatisticalSummary::Statistics )
+
+#endif // QGSSTATISTICALSUMMARY_H

tests/src/core/CMakeLists.txt

@@ -154,3 +154,5 @@ ADD_QGIS_TEST(snappingutilstest testqgssnappingutils.cpp )
 ADD_QGIS_TEST(imageoperationtest testqgsimageoperation.cpp)
 ADD_QGIS_TEST(painteffecttest testqgspainteffect.cpp)
 ADD_QGIS_TEST(painteffectregistrytest testqgspainteffectregistry.cpp)
+ADD_QGIS_TEST(statisticalsummarytest testqgsstatisticalsummary.cpp)
+

tests/src/core/testqgsstatisticalsummary.cpp

@@ -0,0 +1,90 @@
+/***************************************************************************
+     testqgsstatisticalsummary.cpp
+     -----------------------------
+    Date                 : May 2015
+    Copyright            : (C) 2015 Nyall Dawson
+    Email                : nyall dot dawson at gmail dot com
+ ***************************************************************************
+ *                                                                         *
+ *   This program is free software; you can redistribute it and/or modify  *
+ *   it under the terms of the GNU General Public License as published by  *
+ *   the Free Software Foundation; either version 2 of the License, or     *
+ *   (at your option) any later version.                                   *
+ *                                                                         *
+ ***************************************************************************/
+#include <QtTest/QtTest>
+#include <QObject>
+#include <QString>
+#include <QStringList>
+#include <QSettings>
+#include <QSharedPointer>
+
+#include "qgsstatisticalsummary.h"
+#include "qgis.h"
+
+class TestQgsStatisticSummary: public QObject
+{
+    Q_OBJECT
+
+  private slots:
+    void initTestCase();// will be called before the first testfunction is executed.
+    void cleanupTestCase();// will be called after the last testfunction was executed.
+    void init();// will be called before each testfunction is executed.
+    void cleanup();// will be called after every testfunction.
+    void stats();
+
+  private:
+
+};
+
+void TestQgsStatisticSummary::initTestCase()
+{
+
+}
+
+void TestQgsStatisticSummary::cleanupTestCase()
+{
+
+}
+
+void TestQgsStatisticSummary::init()
+{
+
+}
+
+void TestQgsStatisticSummary::cleanup()
+{
+
+}
+
+void TestQgsStatisticSummary::stats()
+{
+  QgsStatisticalSummary s( QgsStatisticalSummary::All );
+  QList<double> values;
+  values << 4 << 2 << 3 << 2 << 5 << 8;
+  s.calculate( values );
+
+  QCOMPARE( s.count(), 6 );
+  QCOMPARE( s.sum(), 24.0 );
+  QCOMPARE( s.mean(), 4.0 );
+  QVERIFY( qgsDoubleNear( s.stDev(), 2.0816, 0.0001 ) );
+  QVERIFY( qgsDoubleNear( s.sampleStDev(), 2.2803, 0.0001 ) );
+
+  QCOMPARE( s.min(), 2.0 );
+  QCOMPARE( s.max(), 8.0 );
+  QCOMPARE( s.range(), 6.0 );
+
+  QCOMPARE( s.median(), 3.5 );
+  values << 9;
+  s.calculate( values );
+  QCOMPARE( s.median(), 4.0 );
+
+  values << 4 << 5 << 8 << 12 << 12 << 12;
+  s.calculate( values );
+  QCOMPARE( s.variety(), 7 );
+  QCOMPARE( s.minority(), 3.0 );
+  QCOMPARE( s.majority(), 12.0 );
+}
+
+QTEST_MAIN( TestQgsStatisticSummary )
+#include "testqgsstatisticalsummary.moc"