Changeset 15926 for branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab

Timestamp:

05/02/18 13:14:16 (7 years ago)

Author:

rhanghof

Message:

#2913:

• Added the support for importing different Matlab datatypes.
• Added some classes and changed the import dialog for importing double arrays.

Location:

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab

Files:

• Api/IMatlabConnector.cs (modified) (3 diffs)
• Api/MatlabConnector.cs (modified) (4 diffs)
• Api/Types/MLDatatype.cs (modified) (1 diff)
• Api/Types/MLDoubleArray.cs (modified) (4 diffs)
• Api/Types/MLTimeseries.cs (modified) (3 diffs)
• Api/Types/MLVariableInfo.cs (added)
• RegressionMatlabInstanceProvider.cs (modified) (1 diff)

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/Api/IMatlabConnector.cs

@@ -9 +9 @@
 namespace HeuristicLab.Problems.Instances.DataAnalysis.Regression.Matlab.Api {
   public interface IMatlabConnector : IDisposable {
+    event Action<string> ErrorOccuredEvent;
+
     /// <summary>
     /// Executes a given matlab script.
@@ -25 +27 @@
     /// </summary>
     /// <returns></returns>
-    IDictionary<string, MLDatatype> GetVariablesFromScript(string script);
+    IEnumerable<MLVariableInfo> GetVariablesFromScript(string script);
 
     /// <summary>
@@ -47 +49 @@
     /// <returns></returns>
     MLDouble GetDouble(string varName);
+
+    /// <summary>
+    /// 
+    /// </summary>
+    /// <param name="varName"></param>
+    /// <returns></returns>
+    MLDoubleArray GetDoubleArray(string varName);
   }
 }

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/Api/MatlabConnector.cs

@@ -31 +31 @@
     private MLApp.MLApp matLabApi;
     private string workspace;
+    private object locker = new object();
+
+    public event Action<string> ErrorOccuredEvent;
 
     public MatlabConnector(bool showcommandWindow = false, string workspace = "base") {
@@ -68 +71 @@
 
 
-    public IDictionary<string, MLDatatype> GetVariablesFromScript(string script) {
-      var variables = new Dictionary<string, MLDatatype>();
-      ExecuteScript(script);
+    public IEnumerable<MLVariableInfo> GetVariablesFromScript(string script) {
+      var variables = new List<MLVariableInfo>();
+      try {
+        ExecuteScript(script);
+      } catch(InvalidOperationException ex) {
+        if (ErrorOccuredEvent != null) {
+          ErrorOccuredEvent.Invoke(ex.Message);
+        }
+      }
 
       Execute("variables=whos();nrOfVariables=size(variables)");
@@ -77 +86 @@
       for (int i = 1; i <= len; i++) {
         var varName = GetVariable(string.Format("variables({0}).name", i)) as string;
-        var varSize = GetVariable(string.Format("variables({0}).size", i));
+        var varSize = GetVariable(string.Format("variables({0}).size", i)) as double[,];
         var varClass = GetVariable(string.Format("variables({0}).class", i)) as string;
-        // At the moment only timeseries are supported.
-        if (varClass == "timeseries") {
-          variables.Add(varName, MLDatatype.Timeseries);
-        }
+        
+        GetVariablesFromScriptRec(variables, varName, varSize, varClass);
       }
 
       return variables;
     }
-    
+
+    private void GetVariablesFromScriptRec(IList<MLVariableInfo> variables, string varName, double[,] varSize, string varClass) {
+      var variable = new MLVariableInfo() {
+        VarName = varName,
+        Row = varSize[0, 0],
+        Column = varSize[0, 1]
+      };
+      if (varClass == "timeseries") {
+        variable.Datatype = MLDatatype.Timeseries;
+        variables.Add(variable);
+      } else if (varClass == "double") {
+        if (variable.Row == 1 && variable.Column == 1) {
+          variable.Datatype = MLDatatype.Double;
+        } else {
+          variable.Datatype = MLDatatype.DoubleArray;
+        }
+        variables.Add(variable);
+      } else if (varClass == "struct") {
+        Execute(string.Format("mlFieldNames = fieldnames({0})", varName));
+        var fields = GetVariable("mlFieldNames") as object[,];
+
+        for (int i = 0; i < fields.GetLength(0); i++) {
+          var newVarName = string.Format("{0}.{1}", varName, fields[i, 0] as string);
+          Execute(string.Format("mlClass = class({0}); mlSize = size({0});", newVarName));
+          var newVarClass = GetVariable("mlClass") as string;
+          var newVarSize = GetVariable("mlSize") as double[,];
+          GetVariablesFromScriptRec(variables, newVarName, newVarSize, newVarClass);
+        }
+      }
+    }
 
     public MLTimeseries GetTimeseries(string varName) {
-      object times = GetVariable(varName + ".Time");
-      object data = GetVariable(varName + ".Data");
-      if (times == null || data == null) {
-        return null;
+      lock (locker) {
+        object times = GetVariable(varName + ".Time");
+        object data = GetVariable(varName + ".Data");
+
+        if (times == null || data == null) {
+          return null;
+        }
+        return new MLTimeseries(varName, times, data);
       }
-
-      return new MLTimeseries(varName, times, data);
     }
 
     public MLDouble GetDouble(string varName) {
       return new MLDouble(varName, GetVariable(varName));
+    }
+
+
+    public MLDoubleArray GetDoubleArray(string varName) {
+      return new MLDoubleArray(varName, GetVariable(varName));
     }
 
@@ -142 +185 @@
       }
     }
-
   }
 }

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/Api/Types/MLDatatype.cs

@@ -9 +9 @@
         Double,
         DoubleArray,
-        Timeseries
+        Timeseries,
+        Struct,
+        Undefinded
     }
 }

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/Api/Types/MLDoubleArray.cs

@@ -27 +27 @@
       Name = string.Format("{0};{1}", a1.Name, a2.Name);
 
-      if (!ConcatFeasible(a1, a2)) {
+      if (!ConcatIsPossible(a1, a2)) {
         throw new ArgumentException(string.Format("Cannot concat {0} and {1}. Different number of rows.", a1.Name, a2.Name));
       }
 
-      var numberOfColumns = a1.Columns + a2.Columns;
-      _dataHeaders = new string[numberOfColumns];
-      for (int i = 0; i < numberOfColumns; i++) {
+      var columns = a1.Columns + a2.Columns;
+      _dataHeaders = new string[columns];
+      for (int i = 0; i < columns; i++) {
         if (i < a1.Columns) {
           _dataHeaders[i] = a1.DataHeader[i];
@@ -41 +41 @@
       }
 
-      Data = new double[a1.Rows, numberOfColumns];
+      Data = new double[a1.Rows, columns];
       for (int i = 0; i < a1.Rows; i++) {
-        for (int j = 0; j < numberOfColumns; j++) {
+        for (int j = 0; j < columns; j++) {
           if (j < a1.Columns) {
             Data[i, j] = a1.Data[i, j];
           } else {
             Data[i, j] = a1.Data[i, j - a1.Columns];
+          }
+        }
+      }
+    }
+
+    private MLDoubleArray(IEnumerable<MLDoubleArray> arrays) {
+      StringBuilder sb = new StringBuilder();
+      foreach (var array in arrays) {
+        sb.Append(array.Name + ";");
+      }
+      Name = sb.ToString();
+
+      var columns = 0;
+      foreach (var array in arrays) {
+        columns += array.Columns;
+      }
+
+      var dataHeaders = new List<string>();
+      foreach (var array in arrays) {
+        foreach (var header in array.DataHeader) {
+          dataHeaders.Add(header);
+        }
+      }
+      _dataHeaders = dataHeaders.ToArray();
+
+      var rows = arrays.First().Rows;
+      Data = new double[rows, columns];
+      for (int i = 0; i < rows; i++) {
+        var j = 0;
+        foreach (var array in arrays) {
+          for (int k = 0; k < array.Columns; k++) {
+            Data[i, j] = array.Data[i, k];
+            j++;
           }
         }
@@ -57 +90 @@
     /// <summary>
     /// Concats two given MLDoubleArrays if their number of rows equals.
-    /// If they can't be concated, this method returns null.
+    /// If throws an InvalidOperationException if they can't be concatenated
     /// </summary>
     /// <param name="a1"></param>
@@ -63 +96 @@
     /// <returns></returns>
     public static MLDoubleArray Concat(MLDoubleArray a1, MLDoubleArray a2) {
-      if (ConcatFeasible(a1, a2)) {
-        return null;
+      if (ConcatIsPossible(a1, a2)) {
+        throw new InvalidOperationException("Concatination is not possible because of different different number of their rows.");
       }
       return new MLDoubleArray(a1, a2);
     }
 
-    public static bool ConcatFeasible(MLDoubleArray a1, MLDoubleArray a2) {
+    public static MLDoubleArray Concat(IEnumerable<MLDoubleArray> arrays) {
+      if (arrays.Count() < 1) {
+        throw new ArgumentException("Given number of MLDoubleArrays is 0");
+      }
+      if (!ConcatIsPossible(arrays)) {
+        throw new InvalidOperationException("Concatenation is not possible because of different different number of their rows.");
+      }
+      return new MLDoubleArray(arrays);
+    }
+
+    public static bool ConcatIsPossible(MLDoubleArray a1, MLDoubleArray a2) {
       return a1.Rows == a2.Rows;
+    }
+
+    public static bool ConcatIsPossible(IEnumerable<MLDoubleArray> arrays) {
+      var first = arrays.First();
+      foreach (var array in arrays) {
+        if (!ConcatIsPossible(first, array)) {
+          return false;
+        }
+      }
+      return true;
     }
 

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/Api/Types/MLTimeseries.cs

@@ -3 +3 @@
 using System.Linq;
 using System.Text;
+using System.Threading;
 using System.Threading.Tasks;
 
@@ -24 +25 @@
     /// </summary>
     /// <param name="timeseries"></param>
-    public MLTimeseries(IEnumerable<MLTimeseries> timeseries) : this() {
+    public MLTimeseries(IEnumerable<MLTimeseries> timeseries, CancellationToken token) : this() {
       var dataHeaders = new List<string>();
       var times = new SortedSet<double>(); // The time values have to be unique and sorted.
@@ -64 +65 @@
 
           foreach (var v in vs) {
+            if (token.IsCancellationRequested) {
+              return;
+            }
             values[idx++] = v;
           }

branches/2913_MatlabScriptProblemInstanceProvider/HeuristicLab.Problems.Instances.DataAnalysis/3.3/Regression/Matlab/RegressionMatlabInstanceProvider.cs

@@ -67 +67 @@
 
 
-    public IRegressionProblemData ImportData(string path, RegressionMatlabImportType type, IEnumerable<KeyValuePair<string, MLDatatype>> variableNames) {
+    public IRegressionProblemData ImportData(string path, RegressionMatlabImportType type, IEnumerable<MLVariableInfo> variableNames) {
       var dataset = type.Values;
       var targetVar = type.TargetVariable;