Changeset 10381 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing

Timestamp:

01/22/14 16:13:42 (11 years ago)

Author:

mleitner

Message:

Add Media, Average, Variance, and StdDeviation support for DateTime

Location:

branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3

Files:

• Implementations/StatisticsLogic.cs (modified) (5 diffs)
• Interfaces/IStatisticsLogic.cs (modified) (2 diffs)

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

@@ -2 +2 @@
 using System.Linq;
 using HeuristicLab.Common;
+using System.Collections.Generic;
 
 namespace HeuristicLab.DataPreprocessing {
@@ -74 +75 @@
     }
 
+    public DateTime GetMedianDateTime(int columnIndex)
+    {
+      DateTime median = new DateTime();
+      if (preprocessingData.IsType<DateTime>(columnIndex))
+      {
+          median = GetSecondsAsDateTime(GetDateTimeAsSeconds(columnIndex).Median());
+      }
+      return median;
+    }
+
+    public DateTime GetAverageDateTime(int columnIndex)
+    {
+      DateTime avg = new DateTime();
+      if (preprocessingData.IsType<DateTime>(columnIndex))
+      {
+          avg = GetSecondsAsDateTime(GetDateTimeAsSeconds(columnIndex).Average());   
+      }
+      return avg;
+    }
+
     public T GetMostCommonValue<T>(int columnIndex) {
       return preprocessingData.GetValues<T>(columnIndex)
+
                               .GroupBy(x => x)
                               .OrderByDescending(g => g.Count())
@@ -88 +110 @@
         stdDev = preprocessingData.GetValues<double>(columnIndex).StandardDeviation();
       }
+      else if (preprocessingData.IsType<DateTime>(variableName)) {
+        stdDev = GetDateTimeAsSeconds(variableName).StandardDeviation();
+      }
       return stdDev;
     }
 
     public double GetVariance(int columnIndex) {
-      double stdDev = double.NaN;
+       double variance = double.NaN;
+
       if (preprocessingData.IsType<double>(columnIndex)) {
-        stdDev = preprocessingData.GetValues<double>(columnIndex).Variance();
+        variance = preprocessingData.GetValues<double>(columnIndex).Variance();
       }
-      return stdDev;
+      else if (preprocessingData.IsType<DateTime>(variableName)) {
+        variance = GetDateTimeAsSeconds(variableName).Variance();
+      }
+      return variance;
     }
 
@@ -112 +141 @@
       return count;
     }
-
 
     public string GetVariableName(int columnIndex) {
@@ -133 +161 @@
     }
 
+    private IEnumerable<double> GetDateTimeAsSeconds(int columnIndex)
+    {
+        return preprocessingData.GetValues<DateTime>(columnIndex).Select(x => (double)x.Ticks / TimeSpan.TicksPerSecond);
+    }
+
+    private DateTime GetSecondsAsDateTime(double seconds)
+    {
+        DateTime dateTime = new DateTime();
+        return dateTime.Add(new TimeSpan(0, 0, (int)seconds));
+    }
   }
 }

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

@@ -11 +11 @@
     int GetMissingValueCount(int columnIndex);
     int GetRowMissingValueCount(int rowIndex);
+
     T GetMin<T>(int columnIndex) where T : IComparable<T>;
     T GetMax<T>(int columnIndex) where T : IComparable<T>;
+    
     double GetMedian(int columnIndex);
     double GetAverage(int columnIndex);
+    DateTime GetMedianDateTime(int columnIndex);
+    DateTime GetAverageDateTime(int columnIndex);
+
     double GetStandardDeviation(int columnIndex);
     double GetVariance(int columnIndex);
@@ -22 +27 @@
     bool IsType<T>(int columnIndex);
     string GetColumnTypeAsString(int columnIndex);
+
   }
 }