Changeset 10367 for branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/StatisticsLogic.cs

Timestamp:

01/22/14 14:13:11 (10 years ago)

Author:

rstoll

Message:

• modified PreprocessingData, uses columnIndex now instead of variableName (is faster and more convenient), set variabelName based methods to Obsolete
• Already changed SearchLogic, DataGridLogic, StatisticLogic as well as PreprocessingDataManipulation

*

File:

• branches/DataPreprocessing/HeuristicLab.DataPreprocessing/3.3/Implementations/StatisticsLogic.cs (modified) (3 diffs)

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

@@ -39 +39 @@
     public int GetMissingValueCount() {
       int count = 0;
-      foreach (var variableName in preprocessingData.VariableNames) {
-        count += GetMissingValueCount(variableName);
+      for(int i = 0; i < preprocessingData.Columns; ++i){
+        count += GetMissingValueCount(i);
       }
       return count;
     }
 
-    public int GetMissingValueCount(string variableName) {
-      return searchLogic.GetMissingValueIndices(variableName).Count();
+    public int GetMissingValueCount(int columnIndex) {
+      return searchLogic.GetMissingValueIndices(columnIndex).Count();
     }
 
-    public T GetMin<T>(string variableName) where T : IComparable<T> {
-      return preprocessingData.GetValues<T>(variableName).Min();
+    public T GetMin<T>(int columnIndex) where T : IComparable<T> {
+      return preprocessingData.GetValues<T>(columnIndex).Min();
     }
 
-    public T GetMax<T>(string variableName) where T : IComparable<T> {
-      return preprocessingData.GetValues<T>(variableName).Max();
+    public T GetMax<T>(int columnIndex) where T : IComparable<T> {
+      return preprocessingData.GetValues<T>(columnIndex).Max();
     }
 
-    public double GetMedian(string variableName) {
+    public double GetMedian(int columnIndex) {
       double median = double.NaN;
-      if (preprocessingData.IsType<double>(variableName)) {
-        median = preprocessingData.GetValues<double>(variableName).Median();
+      if (preprocessingData.IsType<double>(columnIndex)) {
+        median = preprocessingData.GetValues<double>(columnIndex).Median();
       }
       return median;
     }
 
-    public double GetAverage(string variableName) {
+    public double GetAverage(int columnIndex) {
       double avg = double.NaN;
-      if (preprocessingData.IsType<double>(variableName)) {
-        avg = preprocessingData.GetValues<double>(variableName).Average();
+      if (preprocessingData.IsType<double>(columnIndex)) {
+        avg = preprocessingData.GetValues<double>(columnIndex).Average();
       }
       return avg;
     }
 
-    public T GetMostCommonValue<T>(string variableName) {
-      return preprocessingData.GetValues<T>(variableName)
+    public T GetMostCommonValue<T>(int columnIndex) {
+      return preprocessingData.GetValues<T>(columnIndex)
                               .GroupBy(x => x)
                               .OrderByDescending(g => g.Count())
@@ -82 +82 @@
 
 
-    public double GetStandardDeviation(string variableName) {
+    public double GetStandardDeviation(int columnIndex) {
       double stdDev = double.NaN;
-      if (preprocessingData.IsType<double>(variableName)) {
-        stdDev = preprocessingData.GetValues<double>(variableName).StandardDeviation();
+      if (preprocessingData.IsType<double>(columnIndex)) {
+        stdDev = preprocessingData.GetValues<double>(columnIndex).StandardDeviation();
       }
       return stdDev;
     }
 
-    public double GetVariance(string variableName) {
+    public double GetVariance(int columnIndex) {
       double stdDev = double.NaN;
-      if (preprocessingData.IsType<double>(variableName)) {
-        stdDev = preprocessingData.GetValues<double>(variableName).Variance();
+      if (preprocessingData.IsType<double>(columnIndex)) {
+        stdDev = preprocessingData.GetValues<double>(columnIndex).Variance();
       }
       return stdDev;
     }
 
-    public int GetDifferentValuesCount<T>(string variableName) {
-      return preprocessingData.GetValues<T>(variableName).GroupBy(x => x).Count();
+    public int GetDifferentValuesCount<T>(int columnIndex) {
+      return preprocessingData.GetValues<T>(columnIndex).GroupBy(x => x).Count();
     }
 
     public int GetRowMissingValueCount(int rowIndex) {
       int count = 0;
-      foreach (var variableName in preprocessingData.VariableNames) {
-        if (searchLogic.IsMissingValue(variableName, rowIndex)) {
+       for(int i = 0; i < preprocessingData.Columns; ++i){
+        if (searchLogic.IsMissingValue(i, rowIndex)) {
           ++count;
         }
@@ -111 +111 @@
       return count;
     }
+
+
+    public string GetVariableName(int columnIndex) {
+      return preprocessingData.GetVariableName(columnIndex);
+    }
+
+    public bool IsType<T>(int columnIndex) {
+      return preprocessingData.IsType<T>(columnIndex);
+    }
   }
 }