Changeset 14724 for branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4

Timestamp:

03/07/17 15:32:44 (8 years ago)

Author:

mkommend

Message:

#2709: Adapted data preprocessing scatter plot to allow grouping of string variables.

File:

• branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/ScatterPlotContent.cs (modified) (3 diffs)

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

@@ -25 +25 @@
 using HeuristicLab.Analysis;
 using HeuristicLab.Common;
-using HeuristicLab.Problems.DataAnalysis;
 using HeuristicLab.Visualization.ChartControlsExtensions;
 
@@ -55 +54 @@
           scatterPlot.VisualProperties.XAxisMinimumFixedValue = axisMin;
           scatterPlot.VisualProperties.XAxisMaximumFixedValue = axisMax;
-        } catch (ArgumentOutOfRangeException) { } // error during CalculateOptimalAxisInterval 
+        }
+        catch (ArgumentOutOfRangeException) { } // error during CalculateOptimalAxisInterval 
         try {
           double axisMin, axisMax, axisInterval;
@@ -63 +63 @@
           scatterPlot.VisualProperties.YAxisMinimumFixedValue = axisMin;
           scatterPlot.VisualProperties.YAxisMaximumFixedValue = axisMax;
-        } catch (ArgumentOutOfRangeException) { } // error during CalculateOptimalAxisInterval 
+        }
+        catch (ArgumentOutOfRangeException) { } // error during CalculateOptimalAxisInterval 
       }
 
-      if (variableNameGroup == null || variableNameGroup == "-") {
+
+      //No Grouping
+      if (string.IsNullOrEmpty(variableNameGroup) || variableNameGroup == "-") {
         ScatterPlotDataRow scdr = new ScatterPlotDataRow(variableNameX + " - " + variableNameY, "", validPoints);
         scdr.VisualProperties.IsVisibleInLegend = false;
         scatterPlot.Rows.Add(scdr);
-      } else {
-        var groupValues = PreprocessingData.GetValues<double>(PreprocessingData.GetColumnIndex(variableNameGroup));
-        var data = points.Zip(groupValues, (p, g) => new { p, g })
-          .Where(x => !double.IsNaN(x.p.X) && !double.IsNaN(x.p.Y) && !double.IsInfinity(x.p.X) && !double.IsInfinity(x.p.Y))
-          .ToList();
+        return scatterPlot;
+      }
 
-        foreach (var groupValue in groupValues.Distinct().OrderBy(g => g)) {
-          var values = data.Where(x => x.g == groupValue || (double.IsNaN(x.g) && double.IsNaN(groupValue))).Select(v => v.p);
-          var row = new ScatterPlotDataRow(string.Format("{0} ({1})", variableNameGroup, groupValue), "", values) {
-            Name = groupValue.ToString("R"),
-            VisualProperties = { PointSize = 6 }
-          };
-          scatterPlot.Rows.Add(row);
-        }
+      //Grouping
+      int groupVariableIndex = PreprocessingData.GetColumnIndex(variableNameGroup);
+      var groupingValues = Enumerable.Empty<string>();
+
+      if (PreprocessingData.VariableHasType<double>(groupVariableIndex)) {
+        groupingValues = PreprocessingData.GetValues<double>(groupVariableIndex).Select(x => x.ToString());
+      } else if (PreprocessingData.VariableHasType<string>(groupVariableIndex)) {
+        groupingValues = PreprocessingData.GetValues<string>(groupVariableIndex);
+      } else if (PreprocessingData.VariableHasType<DateTime>(groupVariableIndex)) {
+        groupingValues = PreprocessingData.GetValues<DateTime>(groupVariableIndex).Select(x => x.ToString());
+      }
+      var groups = groupingValues.Zip(validPoints, Tuple.Create).GroupBy(t => t.Item1, t => t.Item2);
+
+      foreach (var group in groups) {
+        var scdr = new ScatterPlotDataRow();
+        scdr.Name = group.Key;
+        scdr.VisualProperties.IsVisibleInLegend = true;
+        scdr.VisualProperties.PointSize = 6;
+        scdr.Points.AddRange(group);
+        scatterPlot.Rows.Add(scdr);
       }
       return scatterPlot;
     }
 
-    public DataRow GetCorrelationRow(string variableNameX, string variableNameY) {
-      var xValues = PreprocessingData.GetValues<double>(PreprocessingData.GetColumnIndex(variableNameX));
-      var yValues = PreprocessingData.GetValues<double>(PreprocessingData.GetColumnIndex(variableNameY));
-
-      double k, d;
-      OnlineCalculatorError err;
-      OnlineLinearScalingParameterCalculator.Calculate(xValues, yValues, out k, out d, out err);
-      double p = OnlinePearsonsRCalculator.Calculate(xValues, yValues, out err);
-
-      var data = new double[xValues.Count];
-      for (int i = 0; i < xValues.Count; i++) {
-        data[i]= k * i + d;
-      }
-
-      return new DataRow(string.Format("Correlation (R²={0})", p*p), "", data);
-    }
   }
 }