Changeset 14579

Timestamp:

01/18/17 11:17:19 (7 years ago)

Author:

mkommend

Message:

#2709: Refactored histogram view and content to support grouping by string and datetime variables.

Location:

branches/DataPreprocessing Enhancements

Files:

• HeuristicLab.DataPreprocessing.Views/3.4/HistogramView.cs (modified) (3 diffs)
• HeuristicLab.DataPreprocessing/3.4/Content/HistogramContent.cs (modified) (3 diffs)

branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing.Views/3.4/HistogramView.cs

@@ -21 +21 @@
 
 using System;
-using System.Collections.Generic;
 using System.Linq;
 using HeuristicLab.Analysis;
@@ -27 +26 @@
 
 namespace HeuristicLab.DataPreprocessing.Views {
-
   [View("Histogram View")]
   [Content(typeof(HistogramContent), true)]
   public partial class HistogramView : PreprocessingChartView {
-
-    public List<double> Classification { get; set; }
-    public bool IsDetailedChartViewEnabled { get; set; }
+    public new HistogramContent Content {
+      get { return (HistogramContent)base.Content; }
+      set { base.Content = value; }
+    }
 
     public HistogramView() {
@@ -41 +40 @@
     protected override void OnContentChanged() {
       base.OnContentChanged();
+      classifierComboBox.Items.Clear();
+      classifierComboBox.Items.Add("");
+
       if (Content != null) {
-        classifierComboBox.Items.Clear();
-        classifierComboBox.Items.Add("None");
-
-        foreach (string var in Content.GetVariableNamesForHistogramClassification()) {
+        foreach (string var in Content.GetVariableNamesForGrouping()) {
           classifierComboBox.Items.Add(var);
         }
 
-        if (classifierComboBox.SelectedItem == null && Content.ClassifierVariableIndex < classifierComboBox.Items.Count) {
-          classifierComboBox.SelectedIndex = Content.ClassifierVariableIndex;
-        }
+        classifierComboBox.SelectedItem = Content.GroupingVariableName;
       }
     }
 
     protected override DataTable CreateDataTable(string variableName) {
-      var dt = new DataTable();
-      var row = Content.CreateDataRow(variableName, DataRowVisualProperties.DataRowChartType.Histogram);
-      if (Classification == null) {
-        dt.Rows.Add(row);
-      } else {
-        dt.VisualProperties.Title = variableName;
-        var valuesPerClass = row.Values.Zip(Classification, (value, @class) => new { value, @class })
-                                .GroupBy(x => x.@class)
-                                .ToDictionary(x => x.Key, x => x.Select(v => v.value));
-        foreach (var entry in valuesPerClass) {
-          var classRow = new DataRow(string.Format("{0} ({1})", classifierComboBox.SelectedItem, entry.Key));
-          classRow.VisualProperties.ChartType = DataRowVisualProperties.DataRowChartType.Histogram;
-          classRow.Values.AddRange(entry.Value);
-          dt.Rows.Add(classRow);
-        }
+      var dataTable = new DataTable();
+
+      if (string.IsNullOrEmpty(Content.GroupingVariableName)) {
+        var row = Content.CreateDataRow(variableName, DataRowVisualProperties.DataRowChartType.Histogram);
+        dataTable.Rows.Add(row);
+        return dataTable;
       }
-      return dt;
+
+      dataTable.VisualProperties.Title = variableName;
+
+      int variableIndex = Content.PreprocessingData.GetColumnIndex(variableName);
+      var variableValues = Content.PreprocessingData.GetValues<double>(variableIndex);
+      int groupVariableIndex = Content.PreprocessingData.GetColumnIndex(Content.GroupingVariableName);
+      var groupingValues = Enumerable.Empty<string>();
+
+      if (Content.PreprocessingData.VariableHasType<double>(groupVariableIndex)) {
+        groupingValues = Content.PreprocessingData.GetValues<double>(groupVariableIndex).Select(x => x.ToString());
+      } else if (Content.PreprocessingData.VariableHasType<string>(groupVariableIndex)) {
+        groupingValues = Content.PreprocessingData.GetValues<string>(groupVariableIndex);
+      } else if (Content.PreprocessingData.VariableHasType<DateTime>(groupVariableIndex)) {
+        groupingValues = Content.PreprocessingData.GetValues<DateTime>(groupVariableIndex).Select(x => x.ToString());
+      }
+
+      var groups = groupingValues.Zip(variableValues, Tuple.Create).GroupBy(t => t.Item1, t => t.Item2);
+
+      foreach (var group in groups) {
+        var classRow = new DataRow();
+        classRow.Name = group.Key;
+        classRow.VisualProperties.ChartType = DataRowVisualProperties.DataRowChartType.Histogram;
+        classRow.Values.AddRange(group);
+        dataTable.Rows.Add(classRow);
+      }
+      return dataTable;
     }
 
-    public new HistogramContent Content {
-      get { return (HistogramContent)base.Content; }
-      set { base.Content = value; }
-    }
 
     private void classifierComboBox_SelectedIndexChanged(object sender, EventArgs e) {
-      if (classifierComboBox.SelectedItem == null)
-        return;
-
-      if (classifierComboBox.SelectedIndex != 0) {
-        int columndIndex = Content.PreprocessingData.GetColumnIndex(classifierComboBox.SelectedItem.ToString());
-        Classification = Content.PreprocessingData.GetValues<double>(columndIndex).ToList();
-      } else {
-        Classification = null;
-      }
-
-      Content.ClassifierVariableIndex = classifierComboBox.SelectedIndex;
+      Content.GroupingVariableName = classifierComboBox.SelectedItem.ToString();
 
       // rebuild datatables

branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/HistogramContent.cs

@@ -20 +20 @@
 #endregion
 
+using System;
 using System.Collections.Generic;
 using System.Drawing;
@@ -33 +34 @@
       get { return HeuristicLab.Common.Resources.VSImageLibrary.Statistics; }
     }
-    private const int MAX_DISTINCT_VALUES_FOR_CLASSIFCATION = 20;
 
-    public int ClassifierVariableIndex { get; set; }
+    private const int MAX_DISTINCT_VALUES_FOR_GROUPING = 20;
+
+    public string GroupingVariableName { get; set; }
 
     public int Bins { get; set; }
@@ -53 +55 @@
     }
 
-    public IEnumerable<string> GetVariableNamesForHistogramClassification() {
-      List<string> doubleVariableNames = new List<string>();
+    public IEnumerable<string> GetVariableNamesForGrouping() {
+      var variableNames = new List<string>();
 
-      //only return variable names from type double
       for (int i = 0; i < PreprocessingData.Columns; ++i) {
-        if (PreprocessingData.VariableHasType<double>(i)) {
-          double distinctValueCount = PreprocessingData.GetValues<double>(i).GroupBy(x => x).Count();
-          bool distinctValuesOk = distinctValueCount <= MAX_DISTINCT_VALUES_FOR_CLASSIFCATION;
-          if (distinctValuesOk)
-            doubleVariableNames.Add(PreprocessingData.GetVariableName(i));
-        }
+        int distinctValues = int.MaxValue;
+        if (PreprocessingData.VariableHasType<double>(i))
+          distinctValues = PreprocessingData.GetValues<double>(i).GroupBy(x => x).Count();
+        else if (PreprocessingData.VariableHasType<string>(i))
+          distinctValues = PreprocessingData.GetValues<string>(i).GroupBy(x => x).Count();
+        else if (PreprocessingData.VariableHasType<DateTime>(i))
+          distinctValues = PreprocessingData.GetValues<DateTime>(i).GroupBy(x => x).Count();
+
+        if (distinctValues <= MAX_DISTINCT_VALUES_FOR_GROUPING)
+          variableNames.Add(PreprocessingData.GetVariableName(i));
       }
-      return doubleVariableNames;
+      return variableNames;
     }
   }