Changeset 14725 for branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/PreprocessingChartContent.cs

Timestamp:

03/07/17 16:55:03 (7 years ago)

Author:

mkommend

Message:

#2709: Added grouping for multi scatter plot view.

File:

• branches/DataPreprocessing Enhancements/HeuristicLab.DataPreprocessing/3.4/Content/PreprocessingChartContent.cs (modified) (2 diffs)

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

@@ -66 +66 @@
 
     public DataRow CreateDataRow(string variableName, DataRowVisualProperties.DataRowChartType chartType) {
-      IList<double> values = PreprocessingData.GetValues<double>(PreprocessingData.GetColumnIndex(variableName));
+      return CreateDataRow(PreprocessingData, variableName, chartType);
+    }
+
+    public static DataRow CreateDataRow(IFilteredPreprocessingData preprocessingData, string variableName, DataRowVisualProperties.DataRowChartType chartType) {
+      IList<double> values = preprocessingData.GetValues<double>(preprocessingData.GetColumnIndex(variableName));
       DataRow row = new DataRow(variableName, "", values);
       row.VisualProperties.ChartType = chartType;
@@ -81 +85 @@
       return new ReadOnlyCheckedItemList<StringValue>(itemList);
     }
+
+    private const int MAX_DISTINCT_VALUES_FOR_GROUPING = 20;
+    public static IEnumerable<string> GetVariableNamesForGrouping(IPreprocessingData preprocessingData) {
+      var variableNames = new List<string>();
+
+      for (int i = 0; i < preprocessingData.Columns; ++i) {
+        int distinctValues = Int32.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 variableNames;
+    }
   }
 }