Changeset 10189 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing

Timestamp:

12/04/13 14:18:42 (11 years ago)

Author:

rstoll

Message:

Added GetMissingValueIndices to IPreprocessingData since it will be used in many places
Removed columnIndex specific method from PreprocessingData

Location:

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3

Files:

• Implementations/PreprocessingData.cs (modified) (1 diff)
• Implementations/StatisticInfo.cs (modified) (1 diff)
• Interfaces/IPreprocessingData.cs (modified) (1 diff)

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

@@ -149 +149 @@
     }
 
+    public IDictionary<string, IEnumerable<int>> GetMissingValueIndices() {
+      var dic = new Dictionary<string, IEnumerable<int>>();
+      foreach (string variableName in VariableNames) {
+        dic.Add(variableName, GetMissingValueIndices(variableName));
+      }
+      return dic;
+    }
+
+    public IEnumerable<int> GetMissingValueIndices(string variableName) {
+      if (IsType<double>(variableName)) {
+        return GetValues<double>(variableName).Select((s, i) => new { i, s }).Where(t => double.IsNaN(t.s)).Select(t => t.i);
+      } else if (IsType<string>(variableName)) {
+        return GetValues<string>(variableName).Select((s, i) => new { i, s }).Where(t => string.IsNullOrEmpty(t.s)).Select(t => t.i);
+      } else if (IsType<DateTime>(variableName)) {
+        return GetValues<DateTime>(variableName).Select((s, i) => new { i, s }).Where(t => t.s.Equals(DateTime.MinValue)).Select(t => t.i);
+      } else {
+        throw new ArgumentException("column with index: " + variableName + " contains a non supported type.");
+      }
+    }
+
     #endregion
   }

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

@@ -43 +43 @@
 
     public int GetMissingValueCount(string variableName) {
-      if (preprocessingData.IsType<double>(variableName)) {
-        return preprocessingData.GetValues<double>(variableName).Count(x => double.IsNaN(x));
-      } else if (preprocessingData.IsType<string>(variableName)) {
-        return preprocessingData.GetValues<string>(variableName).Count(x => string.IsNullOrEmpty(x));
-      } else if (preprocessingData.IsType<DateTime>(variableName)) {
-        return preprocessingData.GetValues<DateTime>(variableName).Count(x => x.Equals(DateTime.MinValue));
-      } else {
-        throw new ArgumentException("column with index: " + variableName + " contains a non supported type.");
-      }
+      return preprocessingData.GetMissingValueIndices(variableName).Count();
     }
 

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

@@ -46 +46 @@
 
     void ExportTo(IDataAnalysisProblemData problemData);
+
+    /// <summary>
+    /// Return the indices of the missing values where the key 
+    /// correspdonds to the variable name and the values to the row indices
+    /// </summary>
+    /// <returns></returns>
+    IDictionary<string, IEnumerable<int>> GetMissingValueIndices();
+
+    /// <summary>
+    /// Return the row indices of the cells which contain a missing value
+    /// </summary>
+    /// <returns></returns>
+    IEnumerable<int> GetMissingValueIndices(string variableName);
   }
 }