Add calculation of number of null values to QgsStatisticalSummary

nyalldawson · nyalldawson · commit 1c45b940a83b · 2016-05-17T10:53:04.000+10:00
diff --git a/python/core/qgsstatisticalsummary.sip b/python/core/qgsstatisticalsummary.sip
@@ -22,6 +22,7 @@ class QgsStatisticalSummary
     enum Statistic
     {
       Count,  //!< Count
+      CountMissing, //!< Number of missing (null) values
       Sum,  //!< Sum of values
       Mean,  //!< Mean of values
       Median, //!< Median of values
@@ -79,15 +80,36 @@ class QgsStatisticalSummary
      * @note finalize() must be called after adding the final value and before
      * retrieving calculated statistics.
      * @see calculate()
+     * @see addVariant()
      * @see finalize()
+     * @note added in QGIS 2.16
      */
     void addValue( double value );
 
+    /** Adds a single value to the statistics calculation. Calling this method
+     * allows values to be added to the calculation one at a time. For large
+     * quantities of values this may be more efficient then first adding all the
+     * values to a list and calling calculate().
+     * @param value variant containing to add. Non-numeric values are treated as null.
+     * @note call reset() before adding the first value using this method
+     * to clear the results from any previous calculations
+     * @note finalize() must be called after adding the final value and before
+     * retrieving calculated statistics.
+     * @see addValue()
+     * @see calculate()
+     * @see finalize()
+     * @note added in QGIS 2.16
+     */
+    void addVariant( const QVariant& value );
+
     /** Must be called after adding all values with addValues() and before retrieving
      * any calculated statistics.
      * @see addValue()
+     * @see addVariant()
+     * @note added in QGIS 2.16
      */
     void finalize();
+
     /** Returns the value of a specified statistic
      * @param stat statistic to return
      * @returns calculated value of statistic
diff --git a/src/core/qgsstatisticalsummary.cpp b/src/core/qgsstatisticalsummary.cpp
@@ -39,6 +39,7 @@ QgsStatisticalSummary::~QgsStatisticalSummary()
 void QgsStatisticalSummary::reset()
 {
   mCount = 0;
+  mMissing = 0;
   mSum = 0;
   mMean = 0;
   mMedian = 0;
@@ -88,6 +89,21 @@ void QgsStatisticalSummary::addValue( double value )
     mValues << value;
 }
 
+void QgsStatisticalSummary::addVariant( const QVariant& value )
+{
+  bool convertOk = false;
+  if ( !value.isValid() || value.isNull() )
+    mMissing++;
+  else
+  {
+    double val = value.toDouble( &convertOk );
+    if ( convertOk )
+      addValue( val );
+    else
+      mMissing++;
+  }
+}
+
 void QgsStatisticalSummary::finalize()
 {
   if ( mCount == 0 )
@@ -214,6 +230,8 @@ double QgsStatisticalSummary::statistic( QgsStatisticalSummary::Statistic stat )
   {
     case Count:
       return mCount;
+    case CountMissing:
+      return mMissing;
     case Sum:
       return mSum;
     case Mean:
@@ -254,6 +272,8 @@ QString QgsStatisticalSummary::displayName( QgsStatisticalSummary::Statistic sta
   {
     case Count:
       return QObject::tr( "Count" );
+    case CountMissing:
+      return QObject::tr( "Count (missing)" );
     case Sum:
       return QObject::tr( "Sum" );
     case Mean:
diff --git a/src/core/qgsstatisticalsummary.h b/src/core/qgsstatisticalsummary.h
@@ -17,6 +17,7 @@
 #define QGSSTATISTICALSUMMARY_H
 
 #include <QMap>
+#include <QVariant>
 
 /***************************************************************************
  * This class is considered CRITICAL and any change MUST be accompanied with
@@ -44,6 +45,7 @@ class CORE_EXPORT QgsStatisticalSummary
     enum Statistic
     {
       Count = 1,  //!< Count
+      CountMissing = 32770, //!< Number of missing (null) values
       Sum = 2,  //!< Sum of values
       Mean = 4,  //!< Mean of values
       Median = 8, //!< Median of values
@@ -58,7 +60,7 @@ class CORE_EXPORT QgsStatisticalSummary
       FirstQuartile = 4096, //!< First quartile
       ThirdQuartile = 8192, //!< Third quartile
       InterQuartileRange = 16384, //!< Inter quartile range (IQR)
-      All = Count | Sum | Mean | Median | StDev | Max | Min | Range | Minority | Majority | Variety | FirstQuartile | ThirdQuartile | InterQuartileRange
+      All = Count | CountMissing | Sum | Mean | Median | StDev | Max | Min | Range | Minority | Majority | Variety | FirstQuartile | ThirdQuartile | InterQuartileRange
     };
     Q_DECLARE_FLAGS( Statistics, Statistic )
 
@@ -101,13 +103,33 @@ class CORE_EXPORT QgsStatisticalSummary
      * @note finalize() must be called after adding the final value and before
      * retrieving calculated statistics.
      * @see calculate()
+     * @see addVariant()
      * @see finalize()
+     * @note added in QGIS 2.16
      */
     void addValue( double value );
 
+    /** Adds a single value to the statistics calculation. Calling this method
+     * allows values to be added to the calculation one at a time. For large
+     * quantities of values this may be more efficient then first adding all the
+     * values to a list and calling calculate().
+     * @param value variant containing to add. Non-numeric values are treated as null.
+     * @note call reset() before adding the first value using this method
+     * to clear the results from any previous calculations
+     * @note finalize() must be called after adding the final value and before
+     * retrieving calculated statistics.
+     * @see addValue()
+     * @see calculate()
+     * @see finalize()
+     * @note added in QGIS 2.16
+     */
+    void addVariant( const QVariant& value );
+
     /** Must be called after adding all values with addValues() and before retrieving
      * any calculated statistics.
      * @see addValue()
+     * @see addVariant()
+     * @note added in QGIS 2.16
      */
     void finalize();
 
@@ -121,6 +143,10 @@ class CORE_EXPORT QgsStatisticalSummary
      */
     int count() const { return mCount; }
 
+    /** Returns the number of missing (null) values
+     */
+    int countMissing() const { return mMissing; }
+
     /** Returns calculated sum of values
      */
     double sum() const { return mSum; }
@@ -209,6 +235,7 @@ class CORE_EXPORT QgsStatisticalSummary
     Statistics mStatistics;
 
     int mCount;
+    int mMissing;
     double mSum;
     double mMean;
     double mMedian;
diff --git a/tests/src/core/testqgsstatisticalsummary.cpp b/tests/src/core/testqgsstatisticalsummary.cpp
@@ -35,6 +35,7 @@ class TestQgsStatisticSummary: public QObject
     void individualStatCalculations_data();
     void individualStatCalculations();
     void maxMin();
+    void countMissing();
 
   private:
 
@@ -225,6 +226,7 @@ void TestQgsStatisticSummary::individualStatCalculations_data()
   QTest::newRow( "first_quartile" ) << ( int )QgsStatisticalSummary::FirstQuartile << 3.0;
   QTest::newRow( "third_quartile" ) << ( int )QgsStatisticalSummary::ThirdQuartile << 5.0;
   QTest::newRow( "iqr" ) << ( int )QgsStatisticalSummary::InterQuartileRange << 2.0;
+  QTest::newRow( "missing" ) << ( int )QgsStatisticalSummary::CountMissing << 0.0;
 }
 
 void TestQgsStatisticSummary::individualStatCalculations()
@@ -281,5 +283,21 @@ void TestQgsStatisticSummary::maxMin()
   QCOMPARE( s.max(), -5.0 );
 }
 
+void TestQgsStatisticSummary::countMissing()
+{
+  QgsStatisticalSummary s( QgsStatisticalSummary::All );
+  s.addVariant( 5 );
+  s.addVariant( 6 );
+  s.addVariant( QVariant() );
+  s.addVariant( 7 );
+  s.addVariant( QVariant( QVariant::Double ) );
+  s.addVariant( 9 );
+  s.addVariant( "Asdasdsad" );
+  s.finalize();
+
+  QCOMPARE( s.countMissing(), 3 );
+  QCOMPARE( s.statistic( QgsStatisticalSummary::CountMissing ),  3.0 );
+}
+
 QTEST_MAIN( TestQgsStatisticSummary )
 #include "testqgsstatisticalsummary.moc"
