Changeset 10737

Timestamp:

04/09/14 13:08:49 (10 years ago)

Author:

rstoll

Message:

• Preview and execution for delete columns/rows with insufficient information

Location:

branches/DataPreprocessing

Files:

• HeuristicLab.DataPreprocessing.Views/3.3/HeuristicLab.DataPreprocessing.Views-3.3.csproj (modified) (1 diff)
• HeuristicLab.DataPreprocessing.Views/3.3/ManipulationView.Designer.cs (modified) (11 diffs)
• HeuristicLab.DataPreprocessing.Views/3.3/ManipulationView.cs (modified) (4 diffs)
• HeuristicLab.DataPreprocessing.Views/3.3/ManipulationView.resx (added)
• HeuristicLab.DataPreprocessing/3.3/Implementations/ManipulationLogic.cs (modified) (6 diffs)
• HeuristicLab.DataPreprocessing/3.3/Implementations/SearchLogic.cs (modified) (1 diff)
• HeuristicLab.DataPreprocessing/3.3/Interfaces/IManipulationLogic.cs (modified) (1 diff)

branches/DataPreprocessing/HeuristicLab.DataPreprocessing.Views/3.3/HeuristicLab.DataPreprocessing.Views-3.3.csproj

@@ -243 +243 @@
     </ProjectReference>
   </ItemGroup>
+  <ItemGroup>
+    <EmbeddedResource Include="ManipulationView.resx">
+      <DependentUpon>ManipulationView.cs</DependentUpon>
+    </EmbeddedResource>
+  </ItemGroup>
   <Import Project="$(MSBuildToolsPath)\Microsoft.CSharp.targets" />
   <PropertyGroup>

branches/DataPreprocessing/HeuristicLab.DataPreprocessing.Views/3.3/ManipulationView.Designer.cs

@@ -45 +45 @@
       this.grpBoxPreview = new System.Windows.Forms.GroupBox();
       this.tabsPreview = new System.Windows.Forms.TabControl();
+      this.tabPreviewInactive = new System.Windows.Forms.TabPage();
       this.tabPreviewDeleteColumnsInfo = new System.Windows.Forms.TabPage();
+      this.lblPreviewColumnsInfo = new System.Windows.Forms.Label();
       this.tabPreviewDeleteColumnsVariance = new System.Windows.Forms.TabPage();
+      this.label12 = new System.Windows.Forms.Label();
+      this.lblPreviewColumnsVariance = new System.Windows.Forms.Label();
       this.tabPreviewDeleteRowsInfo = new System.Windows.Forms.TabPage();
+      this.lblPreviewRowsInfo = new System.Windows.Forms.Label();
       this.tabPreviewShuffle = new System.Windows.Forms.TabPage();
-      this.label9 = new System.Windows.Forms.Label();
-      this.tabPreviewInactive = new System.Windows.Forms.TabPage();
+      this.lblPreviewShuffle = new System.Windows.Forms.Label();
       this.grpBoxData.SuspendLayout();
       this.tabsData.SuspendLayout();
@@ -60 +64 @@
       this.grpBoxPreview.SuspendLayout();
       this.tabsPreview.SuspendLayout();
+      this.tabPreviewDeleteColumnsInfo.SuspendLayout();
+      this.tabPreviewDeleteColumnsVariance.SuspendLayout();
+      this.tabPreviewDeleteRowsInfo.SuspendLayout();
       this.tabPreviewShuffle.SuspendLayout();
       this.SuspendLayout();
@@ -194 +201 @@
       // txtDeleteColumnsVariance
       // 
-      this.txtDeleteColumnsVariance.Location = new System.Drawing.Point(226, 0);
+      this.txtDeleteColumnsVariance.Location = new System.Drawing.Point(229, 0);
       this.txtDeleteColumnsVariance.Name = "txtDeleteColumnsVariance";
       this.txtDeleteColumnsVariance.Size = new System.Drawing.Size(135, 20);
@@ -205 +212 @@
       this.label4.Location = new System.Drawing.Point(3, 3);
       this.label4.Name = "label4";
-      this.label4.Size = new System.Drawing.Size(217, 13);
+      this.label4.Size = new System.Drawing.Size(220, 13);
       this.label4.TabIndex = 3;
-      this.label4.Text = "Delete columns with a variance smaller than ";
+      this.label4.Text = "Delete columns with a variance smaller than  ";
       // 
       // tabDataDeleteRowsInfo
@@ -298 +305 @@
       this.tabsPreview.TabIndex = 3;
       // 
+      // tabPreviewInactive
+      // 
+      this.tabPreviewInactive.Location = new System.Drawing.Point(4, 22);
+      this.tabPreviewInactive.Name = "tabPreviewInactive";
+      this.tabPreviewInactive.Padding = new System.Windows.Forms.Padding(3);
+      this.tabPreviewInactive.Size = new System.Drawing.Size(711, 142);
+      this.tabPreviewInactive.TabIndex = 5;
+      this.tabPreviewInactive.Text = "inactive";
+      this.tabPreviewInactive.UseVisualStyleBackColor = true;
+      // 
       // tabPreviewDeleteColumnsInfo
       // 
+      this.tabPreviewDeleteColumnsInfo.Controls.Add(this.lblPreviewColumnsInfo);
       this.tabPreviewDeleteColumnsInfo.Location = new System.Drawing.Point(4, 22);
       this.tabPreviewDeleteColumnsInfo.Name = "tabPreviewDeleteColumnsInfo";
@@ -308 +326 @@
       this.tabPreviewDeleteColumnsInfo.UseVisualStyleBackColor = true;
       // 
+      // lblPreviewColumnsInfo
+      // 
+      this.lblPreviewColumnsInfo.AutoSize = true;
+      this.lblPreviewColumnsInfo.Location = new System.Drawing.Point(3, 3);
+      this.lblPreviewColumnsInfo.Name = "lblPreviewColumnsInfo";
+      this.lblPreviewColumnsInfo.Size = new System.Drawing.Size(318, 13);
+      this.lblPreviewColumnsInfo.TabIndex = 1;
+      this.lblPreviewColumnsInfo.Text = "{0} columns with more than {1} % missing values would be deleted";
+      // 
       // tabPreviewDeleteColumnsVariance
       // 
+      this.tabPreviewDeleteColumnsVariance.Controls.Add(this.label12);
+      this.tabPreviewDeleteColumnsVariance.Controls.Add(this.lblPreviewColumnsVariance);
       this.tabPreviewDeleteColumnsVariance.Location = new System.Drawing.Point(4, 22);
       this.tabPreviewDeleteColumnsVariance.Name = "tabPreviewDeleteColumnsVariance";
@@ -318 +347 @@
       this.tabPreviewDeleteColumnsVariance.UseVisualStyleBackColor = true;
       // 
+      // label12
+      // 
+      this.label12.AutoSize = true;
+      this.label12.Location = new System.Drawing.Point(6, 22);
+      this.label12.Name = "label12";
+      this.label12.Size = new System.Drawing.Size(0, 13);
+      this.label12.TabIndex = 2;
+      // 
+      // lblPreviewColumnsVariance
+      // 
+      this.lblPreviewColumnsVariance.AutoSize = true;
+      this.lblPreviewColumnsVariance.Location = new System.Drawing.Point(3, 3);
+      this.lblPreviewColumnsVariance.Name = "lblPreviewColumnsVariance";
+      this.lblPreviewColumnsVariance.Size = new System.Drawing.Size(301, 13);
+      this.lblPreviewColumnsVariance.TabIndex = 2;
+      this.lblPreviewColumnsVariance.Text = "{0} columns with a variance smaller than {1} would be deleted.";
+      // 
       // tabPreviewDeleteRowsInfo
       // 
+      this.tabPreviewDeleteRowsInfo.Controls.Add(this.lblPreviewRowsInfo);
       this.tabPreviewDeleteRowsInfo.Location = new System.Drawing.Point(4, 22);
       this.tabPreviewDeleteRowsInfo.Name = "tabPreviewDeleteRowsInfo";
@@ -328 +375 @@
       this.tabPreviewDeleteRowsInfo.UseVisualStyleBackColor = true;
       // 
+      // lblPreviewRowsInfo
+      // 
+      this.lblPreviewRowsInfo.AutoSize = true;
+      this.lblPreviewRowsInfo.Location = new System.Drawing.Point(3, 3);
+      this.lblPreviewRowsInfo.Name = "lblPreviewRowsInfo";
+      this.lblPreviewRowsInfo.Size = new System.Drawing.Size(301, 13);
+      this.lblPreviewRowsInfo.TabIndex = 2;
+      this.lblPreviewRowsInfo.Text = "{0} rows with more than {1} % missing values would be deleted";
+      // 
       // tabPreviewShuffle
       // 
-      this.tabPreviewShuffle.Controls.Add(this.label9);
+      this.tabPreviewShuffle.Controls.Add(this.lblPreviewShuffle);
       this.tabPreviewShuffle.Location = new System.Drawing.Point(4, 22);
       this.tabPreviewShuffle.Name = "tabPreviewShuffle";
@@ -339 +395 @@
       this.tabPreviewShuffle.UseVisualStyleBackColor = true;
       // 
-      // label9
-      // 
-      this.label9.AutoSize = true;
-      this.label9.Location = new System.Drawing.Point(3, 3);
-      this.label9.Name = "label9";
-      this.label9.Size = new System.Drawing.Size(252, 13);
-      this.label9.TabIndex = 0;
-      this.label9.Text = "Data will be shuffled randomly - preview not possible";
-      // 
-      // tabPreviewInactive
-      // 
-      this.tabPreviewInactive.Location = new System.Drawing.Point(4, 22);
-      this.tabPreviewInactive.Name = "tabPreviewInactive";
-      this.tabPreviewInactive.Padding = new System.Windows.Forms.Padding(3);
-      this.tabPreviewInactive.Size = new System.Drawing.Size(711, 142);
-      this.tabPreviewInactive.TabIndex = 5;
-      this.tabPreviewInactive.Text = "inactive";
-      this.tabPreviewInactive.UseVisualStyleBackColor = true;
+      // lblPreviewShuffle
+      // 
+      this.lblPreviewShuffle.AutoSize = true;
+      this.lblPreviewShuffle.Location = new System.Drawing.Point(3, 3);
+      this.lblPreviewShuffle.Name = "lblPreviewShuffle";
+      this.lblPreviewShuffle.Size = new System.Drawing.Size(252, 13);
+      this.lblPreviewShuffle.TabIndex = 0;
+      this.lblPreviewShuffle.Text = "Data will be shuffled randomly - preview not possible";
       // 
       // ManipulationView
@@ -382 +428 @@
       this.grpBoxPreview.ResumeLayout(false);
       this.tabsPreview.ResumeLayout(false);
+      this.tabPreviewDeleteColumnsInfo.ResumeLayout(false);
+      this.tabPreviewDeleteColumnsInfo.PerformLayout();
+      this.tabPreviewDeleteColumnsVariance.ResumeLayout(false);
+      this.tabPreviewDeleteColumnsVariance.PerformLayout();
+      this.tabPreviewDeleteRowsInfo.ResumeLayout(false);
+      this.tabPreviewDeleteRowsInfo.PerformLayout();
       this.tabPreviewShuffle.ResumeLayout(false);
       this.tabPreviewShuffle.PerformLayout();
@@ -415 +467 @@
     private System.Windows.Forms.TabPage tabPage1;
     private System.Windows.Forms.Label label7;
-    private System.Windows.Forms.Label label9;
+    private System.Windows.Forms.Label lblPreviewShuffle;
     private System.Windows.Forms.TabPage tabPreviewInactive;
+    private System.Windows.Forms.Label lblPreviewColumnsInfo;
+    private System.Windows.Forms.Label label12;
+    private System.Windows.Forms.Label lblPreviewColumnsVariance;
+    private System.Windows.Forms.Label lblPreviewRowsInfo;
   }
 }

branches/DataPreprocessing/HeuristicLab.DataPreprocessing.Views/3.3/ManipulationView.cs

@@ -45 +45 @@
 
       validators = new Action[] { 
-        ()=>validateDoubleTextBox(txtDeleteColumnsInfo.Text),
-        ()=>validateDoubleTextBox(txtDeleteColumnsVariance.Text),
-        ()=>validateDoubleTextBox(txtDeleteRowsInfo.Text),
-        ()=>{btnApply.Enabled = true;} //shuffle
+        ()=>validateDeleteColumnsInfo(),
+        ()=>validateDeleteColumnsVariance(),
+        ()=>validateDeleteRowsInfo(),
+        ()=>{btnApply.Enabled = true; lblPreviewShuffle.Text = "Data will be shuffled randomly - preview not possible";} //shuffle
       };
+
       manipulations = new Action[] { 
-        ()=>{},
-        ()=>{},
-        ()=>{},
+        ()=>Content.ManipulationLogic.DeleteColumnsWithMissingValuesGreater(getDeleteColumnsInfo()),
+        ()=>Content.ManipulationLogic.DeleteColumnsWithVarianceSmaller(getDeleteColumnsVariance()),
+        ()=>Content.ManipulationLogic.DeleteRowsWithMissingValuesGreater(getRowsColumnsInfo()),
         ()=>Content.ManipulationLogic.ShuffleWithRanges()
       };
+    }
+
+    protected override void RegisterContentEvents() {
+      base.RegisterContentEvents();
+      Content.ManipulationLogic.Changed += Content_Changed;
+    }
+
+    protected override void DeregisterContentEvents() {
+      base.DeregisterContentEvents();
+      Content.ManipulationLogic.Changed -= Content_Changed;
+    }
+
+    private double getDeleteColumnsInfo() {
+      return double.Parse(txtDeleteColumnsInfo.Text);
+    }
+
+    private double getDeleteColumnsVariance() {
+      return double.Parse(txtDeleteColumnsVariance.Text);
+    }
+
+    private double getRowsColumnsInfo() {
+      return double.Parse(txtDeleteRowsInfo.Text);
+    }
+
+    private void validateDeleteColumnsInfo() {
+      validateDoubleTextBox(txtDeleteColumnsInfo.Text);
+      if (btnApply.Enabled) {
+        int count = Content.ManipulationLogic.ColumnsWithMissingValuesGreater(getDeleteColumnsInfo()).Count;
+        lblPreviewColumnsInfo.Text = count + " column" + (count > 1 || count == 0 ? "s" : "") + " were detected with more than " + txtDeleteColumnsInfo.Text + "% missing values.";
+        if (count > 0) {
+          lblPreviewColumnsInfo.Text += Environment.NewLine + Environment.NewLine + "Please press the button \"Apply Manipulation\" if you wish to delete those columns.";
+        } else {
+          btnApply.Enabled = false;
+        }
+      } else {
+        lblPreviewColumnsInfo.Text = "Preview not possible yet - please input the limit above.";
+      }
+    }
+
+    private void validateDeleteColumnsVariance() {
+      validateDoubleTextBox(txtDeleteColumnsVariance.Text);
+      if (btnApply.Enabled) {
+        int count = Content.ManipulationLogic.ColumnsWithVarianceSmaller(getDeleteColumnsVariance()).Count;
+        lblPreviewColumnsVariance.Text = count + " column" + (count > 1 || count == 0 ? "s" : "") + " were detected with a variance smaller than " + txtDeleteColumnsVariance.Text + ".";
+        if (count > 0) {
+          lblPreviewColumnsVariance.Text += Environment.NewLine + Environment.NewLine + "Please press the button \"Apply Manipulation\" if you wish to delete those columns.";
+        } else {
+          btnApply.Enabled = false;
+        }
+      } else {
+        lblPreviewColumnsVariance.Text = "Preview not possible yet - please input the limit for the variance above.";
+      }
+    }
+
+    private void validateDeleteRowsInfo() {
+      validateDoubleTextBox(txtDeleteRowsInfo.Text);
+      if (btnApply.Enabled) {
+        int count = Content.ManipulationLogic.RowsWithMissingValuesGreater(getRowsColumnsInfo()).Count;
+        lblPreviewRowsInfo.Text = count + " row" + (count > 1 || count == 0 ? "s" : "") + " were detected with more than " + txtDeleteRowsInfo.Text + "% missing values.";
+        if (count > 0) {
+          lblPreviewRowsInfo.Text += Environment.NewLine + Environment.NewLine + "Please press the button \"Apply Manipulation\" if you wish to delete those rows.";
+        } else {
+          btnApply.Enabled = false;
+        }
+      } else {
+        lblPreviewRowsInfo.Text = "Preview not possible yet - please input the limit above.";
+      }
     }
 
@@ -70 +138 @@
       btnApply.Enabled = false;
 
-      //in order that button will be enabled if text is already valid
+      //in order that button is enabled if necessary input was already entered
       if (index >= 0) {
         validators[index]();
@@ -78 +146 @@
     private void btnApply_Click(object sender, System.EventArgs e) {
       manipulations[lstMethods.SelectedIndex]();
+      switch (lstMethods.SelectedIndex) {
+        case 0:
+          lblPreviewColumnsInfo.Text = "columns successfully deleted.";
+          break;
+        case 1:
+          lblPreviewColumnsVariance.Text = "columns successfully deleted.";
+          break;
+        case 2:
+          lblPreviewRowsInfo.Text = "rows successfully deleted.";
+          break;
+        case 3:
+          lblPreviewShuffle.Text = "dataset shuffled successfully.";
+          btnApply.Enabled = false;
+          break;
+      }
     }
 
@@ -91 +174 @@
 
     private void txtDeleteColumnsInfo_TextChanged(object sender, EventArgs e) {
-      validateDoubleTextBox(txtDeleteColumnsInfo.Text);
+      validateDeleteColumnsInfo();
     }
 
     private void txtDeleteColumnsVariance_TextChanged(object sender, EventArgs e) {
-      validateDoubleTextBox(txtDeleteColumnsVariance.Text);
+      validateDeleteColumnsVariance();
     }
 
     private void txtDeleteRowsInfo_TextChanged(object sender, EventArgs e) {
-      validateDoubleTextBox(txtDeleteRowsInfo.Text);
+      validateDeleteRowsInfo();
+    }
+
+    void Content_Changed(object sender, DataPreprocessingChangedEventArgs e) {
+      switch(lstMethods.SelectedIndex){
+        case 0:
+          validateDeleteColumnsInfo();
+          break;
+        case 1:
+          validateDeleteColumnsVariance();
+          break;
+        case 2:
+          validateDeleteRowsInfo();
+          break;
+      }
     }
   }

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/ManipulationLogic.cs

@@ -299 +299 @@
       List<int> rows= new List<int>();
 
-      for (int i = preprocessingData.Rows - 1; i >= 0; --i)
+      for (int i = 0; i < preprocessingData.Rows; ++i)
       {
         int missingCount = statisticsLogic.GetRowMissingValueCount(i);
-        if (100f / preprocessingData.Columns * missingCount >= percent)
+        if (100f / preprocessingData.Columns * missingCount > percent)
         {
           rows.Add(i);
@@ -314 +314 @@
 
       List<int> columns = new List<int>();
-      for (int i = preprocessingData.Columns - 1; i >= 0; --i)
-      {
+      for (int i = 0; i < preprocessingData.Columns; ++i) {
         int missingCount = statisticsLogic.GetMissingValueCount(i);
-        if (100f / preprocessingData.Columns * missingCount >= percent)
-        {
+        if (100f / preprocessingData.Rows * missingCount > percent) {
           columns.Add(i);
         }
@@ -329 +327 @@
 
       List<int> columns = new List<int>();
-      for (int i = preprocessingData.Columns - 1; i >= 0; --i)
-      {
+      for (int i = 0; i < preprocessingData.Columns; ++i) {
         if (preprocessingData.IsType<double>(i) || preprocessingData.IsType<DateTime>(i))
         {
@@ -355 +352 @@
     }
 
-    private void DeleteRows(IEnumerable<int> rows) {
+    private void DeleteRows(List<int> rows) {
+      rows.Sort();
+      rows.Reverse();
       preprocessingData.InTransaction(() =>
       {
@@ -365 +364 @@
     }
 
-    private void DeleteColumns(IEnumerable<int> columns) {
+    private void DeleteColumns(List<int> columns) {
+      columns.Sort();
+      columns.Reverse();
       preprocessingData.InTransaction(() =>
       {
@@ -373 +374 @@
         }
       });
+    }
+
+    public event DataPreprocessingChangedEventHandler Changed {
+      add { dataGridLogic.Changed += value; }
+      remove { dataGridLogic.Changed -= value; }
     }
   }

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/SearchLogic.cs

@@ -43 +43 @@
     void preprocessingData_Changed(object sender, DataPreprocessingChangedEventArgs e)
     {
-      MissingValueIndicies.Remove(e.Column);
-      ValuesWithoutNaN.Remove(e.Column);
+      switch (e.Type) {
+        case DataPreprocessingChangedEventType.DeleteColumn:
+        case DataPreprocessingChangedEventType.ChangeColumn:
+          MissingValueIndicies.Remove(e.Column);
+          ValuesWithoutNaN.Remove(e.Column);
+          break;
+        case DataPreprocessingChangedEventType.AddColumn:
+          //cache does not need to be updated, will be calculated the first time it is requested
+          break;
+        case DataPreprocessingChangedEventType.DeleteRow:
+        case DataPreprocessingChangedEventType.AddRow:
+        case DataPreprocessingChangedEventType.ChangeItem:
+        case DataPreprocessingChangedEventType.Any:
+        case DataPreprocessingChangedEventType.Transformation:
+          MissingValueIndicies = new Dictionary<int, IEnumerable<int>>();
+          ValuesWithoutNaN = new Dictionary<int, IEnumerable>();
+          break;
+      } 
     }
 

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Interfaces/IManipulationLogic.cs

@@ -43 +43 @@
     void DeleteColumnsWithMissingValuesGreater(double percent);
     void DeleteColumnsWithVarianceSmaller(double variance);
+
+    event DataPreprocessingChangedEventHandler Changed;
   }
 }