Changeset 17136 for branches/2994-AutoDiffForIntervals/HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Extensions

Timestamp:

07/13/19 10:04:28 (5 years ago)

Author:

gkronber

Message:

#2994: return a bad quality if constraints are violated

File:

• branches/2994-AutoDiffForIntervals/HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Extensions/ConstrainedConstantOptimizationEvaluator.cs (modified) (8 diffs)

branches/2994-AutoDiffForIntervals/HeuristicLab.Problems.DataAnalysis.Regression.Symbolic.Extensions/ConstrainedConstantOptimizationEvaluator.cs

@@ -153 +153 @@
 
         if (ConstantOptimizationRowsPercentage.Value != RelativeNumberOfEvaluatedSamplesParameter.ActualValue.Value) {
-          var evaluationRows = GenerateRowsToEvaluate();
-          quality = SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, double.MinValue, double.MaxValue, ProblemDataParameter.ActualValue, evaluationRows, applyLinearScaling: false);
+          throw new NotSupportedException();
         }
 
@@ -165 +164 @@
 
       } else {
-        var evaluationRows = GenerateRowsToEvaluate();
-        quality = SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(SymbolicDataAnalysisTreeInterpreterParameter.ActualValue, solution, double.MinValue, double.MaxValue, ProblemDataParameter.ActualValue, evaluationRows, applyLinearScaling: false);
+        throw new NotSupportedException();
       }
       QualityParameter.ActualValue = new DoubleValue(quality);
@@ -292 +290 @@
       // buffers for calculate_jacobian
       var target = problemData.TargetVariableTrainingValues.ToArray();
+      var targetVariance = target.VariancePop();
       var fi_eval = new double[target.Length];
       var jac_eval = new double[target.Length, thetaValues.Count];
@@ -337 +336 @@
         try {
           alglib.minnscreate(thetaValues.Count, thetaValues.ToArray(), out state);
-          // alglib.minnssetalgoslp(state);        // SLP is more robust but slower
           alglib.minnssetbc(state, thetaValues.Select(_ => -10000.0).ToArray(), thetaValues.Select(_ => +10000.0).ToArray());
           alglib.minnssetcond(state, 1E-7, maxIterations);
@@ -353 +351 @@
           counter.GradientEvaluations += rep.nfev;
 
-          if (rep.terminationtype != -8) {
+          if (rep.terminationtype > 0) {
             // update parameters in tree
             var pIdx = 0;
@@ -363 +361 @@
               }
             }
-
             // note: we keep the optimized constants even when the tree is worse.
           }
-
+          if (Math.Abs(rep.nlcerr) > 0.01) return targetVariance; // constraints are violated
         } catch (ArithmeticException) {
-          // eval MSE of original tree
-          return SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
-
+          return targetVariance;
         } catch (alglib.alglibexception) {
           // eval MSE of original tree
-          return SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
+          return targetVariance;
         }
       } else if (solver.Contains("minnlc")) {
@@ -413 +408 @@
             // note: we keep the optimized constants even when the tree is worse.
           }
+          if (Math.Abs(rep.nlcerr) > 0.01) return targetVariance; // constraints are violated
 
         } catch (ArithmeticException) {
-          // eval MSE of original tree
-          return SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
-
+          return targetVariance;
         } catch (alglib.alglibexception) {
-          // eval MSE of original tree
-          return SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
+          return targetVariance;
         }
       } else {
@@ -428 +421 @@
 
       // evaluate tree with updated constants
-      return SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
+      var residualVariance = SymbolicRegressionSingleObjectiveMeanSquaredErrorEvaluator.Calculate(interpreter, tree, lowerEstimationLimit, upperEstimationLimit, problemData, rows, applyLinearScaling: false);
+      return Math.Min(residualVariance, targetVariance);
     }