Changeset 15269 for branches/DataPreprocessing Cleanup

Timestamp:

07/18/17 14:19:29 (7 years ago)

Author:

pfleck

Message:

#2809: Removed SearchLogic

Location:

branches/DataPreprocessing Cleanup

Files:

• HeuristicLab.DataPreprocessing.Views/3.4/DataCompletenessView.cs (modified) (2 diffs)
• HeuristicLab.DataPreprocessing.Views/3.4/DataPreprocessingView.cs (modified) (2 diffs)
• HeuristicLab.DataPreprocessing.Views/3.4/ManipulationView.cs (modified) (4 diffs)
• HeuristicLab.DataPreprocessing.Views/3.4/ViewShortcutListView.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Content/DataCompletenessChartContent.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Content/ManipulationContent.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Data/FilteredPreprocessingData.cs (modified) (3 diffs)
• HeuristicLab.DataPreprocessing/3.4/Data/IPreprocessingData.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Data/PreprocessingData.cs (modified) (2 diffs)
• HeuristicLab.DataPreprocessing/3.4/Data/TransactionalPreprocessingData.cs (modified) (2 diffs)
• HeuristicLab.DataPreprocessing/3.4/HeuristicLab.DataPreprocessing-3.4.csproj (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Logic/FilterLogic.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.4/Logic/ManipulationLogic.cs (modified) (4 diffs)
• HeuristicLab.DataPreprocessing/3.4/Logic/SearchLogic.cs (deleted)
• HeuristicLab.DataPreprocessing/3.4/Logic/StatisticsLogic.cs (modified) (6 diffs)

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing.Views/3.4/DataCompletenessView.cs

@@ -52 +52 @@
 
     private void InitData() {
-      IDictionary<int, IList<int>> missingValueIndices = Content.SearchLogic.GetMissingValueIndices();
-
-      bool[,] valueMissing = new bool[Content.SearchLogic.Rows, Content.SearchLogic.Columns];
-      foreach (var columnMissingValues in missingValueIndices) {
-        var column = columnMissingValues.Key;
-        foreach (var missingValueIndex in columnMissingValues.Value)
-          valueMissing[missingValueIndex, column] = true;
+      bool[,] valueMissing = new bool[Content.PreprocessingData.Rows, Content.PreprocessingData.Columns];
+      for (int row = 0; row < Content.PreprocessingData.Rows; row++) {
+        for (int column = 0; column < Content.PreprocessingData.Columns; column++)
+          valueMissing[row, column] = Content.PreprocessingData.IsCellEmpty(column, row);
       }
 
@@ -78 +75 @@
       //custom x axis label
       double from = 0.5;
-      foreach (String columnName in Content.SearchLogic.VariableNames) {
+      foreach (String columnName in Content.PreprocessingData.VariableNames) {
         double to = from + 1;
         chart.ChartAreas[0].AxisX.CustomLabels.Add(from, to, columnName);

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing.Views/3.4/DataPreprocessingView.cs

@@ -50 +50 @@
         var data = Content.Data;
         var filterLogic = new FilterLogic(data);
-        var searchLogic = new SearchLogic(data, filterLogic);
-        var statisticsLogic = new StatisticsLogic(data, searchLogic);
-        var manipulationLogic = new ManipulationLogic(data, searchLogic, statisticsLogic);
+        var statisticsLogic = new StatisticsLogic(data);
+        var manipulationLogic = new ManipulationLogic(data, statisticsLogic);
 
         var viewShortcuts = new ItemList<IViewShortcut> {
@@ -63 +62 @@
           new MultiScatterPlotContent(data),
           new CorrelationMatrixContent(Content),
-          new DataCompletenessChartContent(searchLogic),
+          new DataCompletenessChartContent(data),
 
           new FilterContent(filterLogic),
-          new ManipulationContent(manipulationLogic, searchLogic, filterLogic),
+          new ManipulationContent(manipulationLogic, filterLogic),
           new TransformationContent(data, filterLogic)
         };

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing.Views/3.4/ManipulationView.cs

@@ -127 +127 @@
 
     private void ReplaceMissingValues() {
-      var allIndices = Content.SearchLogic.GetMissingValueIndices();
+      var missingValuesIndices = new List<int>();
       var columnIndex = cmbVariableNames.SelectedIndex;
-      var columnIndices = new Dictionary<int, IList<int>>{
-          {columnIndex,   allIndices[columnIndex]}
-      };
+      var columnIndices = new Dictionary<int, IList<int>> { { columnIndex, missingValuesIndices } };
+
+      for (int rowIndex = 0; rowIndex < Content.ManipulationLogic.PreProcessingData.Rows; rowIndex++)
+        if (Content.ManipulationLogic.PreProcessingData.IsCellEmpty(columnIndex, rowIndex))
+          missingValuesIndices.Add(rowIndex);
 
       switch (cmbReplaceWith.SelectedIndex) {
@@ -165 +167 @@
           sb.Append(Environment.NewLine);
           sb.Append("Columns: ");
-          sb.Append(Content.SearchLogic.VariableNames.ElementAt(filteredColumns.ElementAt(0)));
+          sb.Append(Content.ManipulationLogic.VariableNames.ElementAt(filteredColumns.ElementAt(0)));
           for (int i = 1; i < filteredColumns.Count; i++) {
-            string columnName = Content.SearchLogic.VariableNames.ElementAt(filteredColumns.ElementAt(i));
+            string columnName = Content.ManipulationLogic.VariableNames.ElementAt(filteredColumns.ElementAt(i));
             sb.Append(", ");
             sb.Append(columnName);
@@ -196 +198 @@
           sb.Append(Environment.NewLine);
           sb.Append("Columns: ");
-          sb.Append(Content.SearchLogic.VariableNames.ElementAt(filteredColumns.ElementAt(0)));
+          sb.Append(Content.ManipulationLogic.VariableNames.ElementAt(filteredColumns.ElementAt(0)));
           for (int i = 1; i < filteredColumns.Count; i++) {
-            string columnName = Content.SearchLogic.VariableNames.ElementAt(filteredColumns.ElementAt(i));
+            string columnName = Content.ManipulationLogic.VariableNames.ElementAt(filteredColumns.ElementAt(i));
             sb.Append(", ");
             sb.Append(columnName);
@@ -250 +252 @@
       }
       if (btnApply.Enabled) {
-        var allIndices = Content.SearchLogic.GetMissingValueIndices();
-        int count = allIndices[columnIndex].Count;
+        int count = 0;
+        for (int rowIndex = 0; rowIndex < Content.ManipulationLogic.PreProcessingData.Rows; rowIndex++)
+          if (Content.ManipulationLogic.PreProcessingData.IsCellEmpty(columnIndex, rowIndex)) count++;
+
         int cellCount = Content.FilterLogic.PreprocessingData.Rows * Content.FilterLogic.PreprocessingData.Columns;
         lblPreviewReplaceMissingValues.Text = count + " cell" + (count > 1 || count == 0 ? "s" : "")

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing.Views/3.4/ViewShortcutListView.cs

@@ -42 +42 @@
     //Clone chart items
     protected override void itemsListView_DoubleClick(object sender, EventArgs e) {
-      if (itemsListView.SelectedItems.Count == 1) {
-        IViewShortcut item = itemsListView.SelectedItems[0].Tag as IViewShortcut;
-        if (item != null) {
-          try {
-            item = (IViewShortcut)item.Clone();
-            var view = MainFormManager.MainForm.ShowContent(item);
-            if (view != null) {
-              view.ReadOnly = ReadOnly;
-              view.Locked = Locked;
-            }
-          } catch (NullReferenceException) {
-            // cloning for preprocessing not done properly yet
-          }
+      if (itemsListView.SelectedItems.Count != 1) return;
+      IViewShortcut item = itemsListView.SelectedItems[0].Tag as IViewShortcut;
+      if (item == null) return;
+      try {
+        item = (IViewShortcut)item.Clone();
+        var view = MainFormManager.MainForm.ShowContent(item);
+        if (view != null) {
+          view.ReadOnly = ReadOnly;
+          view.Locked = Locked;
         }
+      } catch (NullReferenceException) {
+        // cloning for preprocessing not done properly yet
       }
     }

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Content/DataCompletenessChartContent.cs

@@ -32 +32 @@
     }
 
-    public SearchLogic SearchLogic { get; private set; }
+    public ITransactionalPreprocessingData PreprocessingData { get; private set; }
 
-    public DataCompletenessChartContent(SearchLogic searchLogic) {
-      SearchLogic = searchLogic;
+    public DataCompletenessChartContent(ITransactionalPreprocessingData preprocessingData) {
+      PreprocessingData = preprocessingData;
     }
 
     public DataCompletenessChartContent(DataCompletenessChartContent content, Cloner cloner)
       : base(content, cloner) {
-      SearchLogic = content.SearchLogic;
+      PreprocessingData = content.PreprocessingData;
     }
 

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Content/ManipulationContent.cs

@@ -33 +33 @@
 
     public ManipulationLogic ManipulationLogic { get; private set; }
-    public SearchLogic SearchLogic { get; private set; }
     public FilterLogic FilterLogic { get; private set; }
 
-    public ManipulationContent(ManipulationLogic manipulationLogic, SearchLogic searchLogic, FilterLogic filterLogic) {
+    public ManipulationContent(ManipulationLogic manipulationLogic, FilterLogic filterLogic) {
       ManipulationLogic = manipulationLogic;
-      SearchLogic = searchLogic;
       FilterLogic = filterLogic;
     }

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Data/FilteredPreprocessingData.cs

@@ -28 +28 @@
 
 namespace HeuristicLab.DataPreprocessing {
-  public class FilteredPreprocessingData : NamedItem, IFilteredPreprocessingData {
+  public sealed class FilteredPreprocessingData : NamedItem, IFilteredPreprocessingData {
     private readonly ITransactionalPreprocessingData originalData;
     private ITransactionalPreprocessingData filteredData;
@@ -83 +83 @@
     }
 
-    protected FilteredPreprocessingData(FilteredPreprocessingData original, Cloner cloner)
+    private FilteredPreprocessingData(FilteredPreprocessingData original, Cloner cloner)
       : base(original, cloner) {
       originalData = original.originalData;
@@ -90 +90 @@
     public override IDeepCloneable Clone(Cloner cloner) {
       return new FilteredPreprocessingData(this, cloner);
+    }
+
+    public bool IsCellEmpty(int columnIndex, int rowIndex) {
+      return ActiveData.IsCellEmpty(columnIndex, rowIndex);
     }
 

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Data/IPreprocessingData.cs

@@ -28 +28 @@
 namespace HeuristicLab.DataPreprocessing {
   public interface IPreprocessingData : INamedItem {
+    bool IsCellEmpty(int columnIndex, int rowIndex);
     T GetCell<T>(int columnIndex, int rowIndex);
 

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Data/PreprocessingData.cs

@@ -160 +160 @@
     }
 
+    public static bool IsMissingValue(object value) {
+      if (value is double) return double.IsNaN((double)value);
+      if (value is string) return string.IsNullOrEmpty((string)value);
+      if (value is DateTime) return ((DateTime)value).Equals(DateTime.MinValue);
+      throw new ArgumentException();
+    }
 
     #region IPreprocessingData Members
+    public abstract bool IsCellEmpty(int columnIndex, int rowIndex);
     public abstract T GetCell<T>(int columnIndex, int rowIndex);
 
@@ -173 +180 @@
 
     public abstract bool VariableHasType<T>(int columnIndex);
-
-    [Obsolete("use the index based variant, is faster")]
-    public abstract IList<T> GetValues<T>(string variableName, bool considerSelection);
 
     public abstract IList<T> GetValues<T>(int columnIndex, bool considerSelection);

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Data/TransactionalPreprocessingData.cs

@@ -88 +88 @@
 
     #region Overridden IPreprocessingData Members
+    public override bool IsCellEmpty(int columnIndex, int rowIndex) {
+      var value = variableValues[columnIndex][rowIndex];
+      return IsMissingValue(value);
+    }
+
     public override T GetCell<T>(int columnIndex, int rowIndex) {
       return (T)variableValues[columnIndex][rowIndex];
@@ -119 +124 @@
     public override bool VariableHasType<T>(int columnIndex) {
       return columnIndex >= variableValues.Count || variableValues[columnIndex] is List<T>;
-    }
-
-    [Obsolete("use the index based variant, is faster")]
-    public override IList<T> GetValues<T>(string variableName, bool considerSelection) {
-      return GetValues<T>(GetColumnIndex(variableName), considerSelection);
     }
 

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/HeuristicLab.DataPreprocessing-3.4.csproj

@@ -146 +146 @@
     <Compile Include="PreprocessingContext.cs" />
     <Compile Include="Data\TransactionalPreprocessingData.cs" />
-    <Compile Include="Logic\SearchLogic.cs" />
     <Compile Include="Logic\StatisticsLogic.cs" />
     <Compile Include="Data\ITransactionalPreprocessingData.cs" />

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Logic/FilterLogic.cs

@@ -73 +73 @@
     }
 
-    public bool Result(bool current, bool addition, bool isAndCombination) {
+    private bool Result(bool current, bool addition, bool isAndCombination) {
       return isAndCombination ? current || addition : current && addition;
     }

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Logic/ManipulationLogic.cs

@@ -30 +30 @@
     private readonly ITransactionalPreprocessingData preprocessingData;
     private readonly StatisticsLogic statisticsLogic;
-    private readonly SearchLogic searchLogic;
 
     public IEnumerable<string> VariableNames {
@@ -40 +39 @@
     }
 
-    public ManipulationLogic(ITransactionalPreprocessingData preprocessingData, SearchLogic theSearchLogic, StatisticsLogic theStatisticsLogic) {
+    public ManipulationLogic(ITransactionalPreprocessingData preprocessingData, StatisticsLogic theStatisticsLogic) {
       this.preprocessingData = preprocessingData;
-      searchLogic = theSearchLogic;
       statisticsLogic = theStatisticsLogic;
     }
@@ -187 +185 @@
     private int indexOfPrevPresentValue(int columnIndex, int start) {
       int offset = start - 1;
-      while (offset >= 0 && searchLogic.IsMissingValue(columnIndex, offset)) {
+      while (offset >= 0 && preprocessingData.IsCellEmpty(columnIndex, offset)) {
         offset--;
       }
@@ -196 +194 @@
     private int indexOfNextPresentValue(int columnIndex, int start) {
       int offset = start + 1;
-      while (offset < preprocessingData.Rows && searchLogic.IsMissingValue(columnIndex, offset)) {
+      while (offset < preprocessingData.Rows && preprocessingData.IsCellEmpty(columnIndex, offset)) {
         offset++;
       }

branches/DataPreprocessing Cleanup/HeuristicLab.DataPreprocessing/3.4/Logic/StatisticsLogic.cs

@@ -28 +28 @@
   public class StatisticsLogic {
     private readonly ITransactionalPreprocessingData preprocessingData;
-    private readonly SearchLogic searchLogic;
-
-    public StatisticsLogic(ITransactionalPreprocessingData thePreprocessingData, SearchLogic theSearchLogic) {
-      preprocessingData = thePreprocessingData;
-      searchLogic = theSearchLogic;
+
+    public StatisticsLogic(ITransactionalPreprocessingData preprocessingData) {
+      this.preprocessingData = preprocessingData;
     }
 
     public int GetColumnCount() {
-      return searchLogic.Columns;
+      return preprocessingData.Columns;
     }
 
     public int GetRowCount() {
-      return searchLogic.Rows;
+      return preprocessingData.Rows;
     }
 
@@ -46 +44 @@
       int count = 0;
 
-      for (int i = 0; i < searchLogic.Columns; ++i) {
+      for (int i = 0; i < preprocessingData.Columns; ++i) {
         if (preprocessingData.VariableHasType<double>(i)) {
           ++count;
@@ -55 +53 @@
 
     public int GetNominalColumnCount() {
-      return searchLogic.Columns - GetNumericColumnCount();
+      return preprocessingData.Columns - GetNumericColumnCount();
     }
 
     public int GetMissingValueCount() {
       int count = 0;
-      for (int i = 0; i < searchLogic.Columns; ++i) {
+      for (int i = 0; i < preprocessingData.Columns; ++i) {
         count += GetMissingValueCount(i);
       }
@@ -67 +65 @@
 
     public int GetMissingValueCount(int columnIndex) {
-      return searchLogic.GetMissingValueIndices(columnIndex).Count();
+      return Enumerable.Range(0, preprocessingData.Rows).Count(rowIndex => preprocessingData.IsCellEmpty(columnIndex, rowIndex));
+    }
+
+    public int GetRowMissingValueCount(int rowIndex) {
+      var columnIndexes = Enumerable.Range(0, preprocessingData.Columns);
+      return columnIndexes.Count(columnIndex => !preprocessingData.IsCellEmpty(columnIndex, rowIndex));
+    }
+
+    private IEnumerable<T> GetValuesWithoutNaN<T>(int columnIndex, bool considerSelection = false) {
+      return preprocessingData.GetValues<T>(columnIndex, considerSelection).Where(x => !PreprocessingData.IsMissingValue(x)).ToList();
     }
 
@@ -185 +192 @@
     }
 
-    public int GetRowMissingValueCount(int rowIndex) {
-      int count = 0;
-      for (int i = 0; i < preprocessingData.Columns; ++i) {
-        if (searchLogic.IsMissingValue(i, rowIndex)) {
-          ++count;
-        }
-      }
-      return count;
-    }
-
-    public string GetVariableName(int columnIndex) {
-      return preprocessingData.GetVariableName(columnIndex);
-    }
+
 
     public bool VariableHasType<T>(int columnIndex) {
@@ -218 +213 @@
     }
 
-    private IEnumerable<T> GetValuesWithoutNaN<T>(int columnIndex, bool considerSelection = false) {
-      return searchLogic.GetValuesWithoutNaN<T>(columnIndex, considerSelection);
-    }
 
     private DateTime GetSecondsAsDateTime(double seconds) {