WIP step 6

Nettoyage du code Repport de la correction Synchornisation avec Carine pour la classe KWDataGridOptimizer
KhiopsML · Oct 18, 2023 · ecb8361 · ecb8361
1 parent 78adba2
commit ecb8361
Show file tree

Hide file tree

Showing 5 changed files with 41 additions and 136 deletions.
diff --git a/src/Learning/KWDataPreparation/KWDataGridOptimizer.cpp b/src/Learning/KWDataPreparation/KWDataGridOptimizer.cpp
@@ -1666,10 +1666,10 @@ double KWDataGridVNSOptimizer::VNSOptimizeDataGrid(const KWDataGrid* initialData
 // CH IV Begin
 double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid* initialDataGrid,
 							      KWDataGridMerger* neighbourDataGrid,
-							      double dNeighbourDataGridCost, KWDataGrid* mergedDataGrid,
+							      double& dNeighbourDataGridCost,
+							      KWDataGrid* mergedDataGrid,
 							      KWDataGrid* partitionedReferencePostMergedDataGrid) const
 {
-	double dCost;
 	double dMergedCost;
 	double dFusionDeltaCost;
 	KWDataGridManager dataGridManager;
@@ -1682,9 +1682,7 @@ double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid*
 	require(neighbourDataGrid->GetDataGridCosts() == dataGridCosts);
 	require(fabs(dNeighbourDataGridCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
 
-	// Initialisations
-	dataGridManager.SetSourceDataGrid(initialDataGrid);
-	dCost = dNeighbourDataGridCost;
+	// Initialisation du cout
 	dMergedCost = dNeighbourDataGridCost;
 
 	// CH AB AF a voir avec Marc
@@ -1698,28 +1696,30 @@ double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid*
 		// Tri des attributs
 		neighbourDataGrid->SortAttributeParts();
 
+		// Initialisation du datagrid manager
 		dataGridManager.SetSourceDataGrid(neighbourDataGrid);
+
 		// Creation d'une nouvelle grille avec nouvelle description des PV fusionnees
 		dFusionDeltaCost =
 		    dataGridManager.ExportDataGridWithVarPartMergeOptimization(mergedDataGrid, dataGridCosts);
 		assert(not mergedDataGrid->GetVarPartsShared());
 
 		// Calcul et verification du cout
-		dMergedCost = dCost + dFusionDeltaCost;
+		dMergedCost = dNeighbourDataGridCost + dFusionDeltaCost;
 
 		// Cas ou le cout de la grille avec PV voisines fusionnees est plus eleve que le cout avant
 		// fusion
-		if (dMergedCost > dCost * (1 + dEpsilon) and bDisplayResults)
+		if (dMergedCost > dNeighbourDataGridCost * (1 + dEpsilon) and bDisplayResults)
 		{
 			sLabel = "PROBLEME : degradation du cout lors de la fusion des parties de variables "
 				 "contigues";
 			sLabel += DoubleToString(dMergedCost);
 			sLabel += "\t";
-			sLabel += DoubleToString(dMergedCost - dCost);
+			sLabel += DoubleToString(dMergedCost - dNeighbourDataGridCost);
 			AddWarning(sLabel);
 			cout << "PROBLEME : degradation du cout lors de la fusion des parties de variables "
 				"contigues : cout fusionne\t "
-			     << dMergedCost << "\tcout\t" << dCost << "\n";
+			     << dMergedCost << "\tcout\t" << dNeighbourDataGridCost << "\n";
 			cout << "Grille avant fusion\n";
 			neighbourDataGrid->Write(cout);
 			cout << "Grille apres fusion\n";
@@ -1792,14 +1792,14 @@ double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid*
 				}
 
 				// Parametrage du profiling
-				KWDataGridOptimizer::GetProfiler()->BeginMethod("Post-optimization IV");
-				KWDataGridOptimizer::GetProfiler()->WriteKeyString("Improvement number",
-										   IntToString(nImprovementNumber));
+				NEWKWDataGridOptimizer::GetProfiler()->BeginMethod("Post-optimization IV");
+				NEWKWDataGridOptimizer::GetProfiler()->WriteKeyString("Improvement number",
+										      IntToString(nImprovementNumber));
 
 				// Exploration des deplacements pour tous les attributs
 				bImprovement = varPartDataGridPostOptimizer.PostOptimizeLightVarPartDataGrid(
 				    partitionedReferencePostMergedDataGrid, mergedDataGrid, &ivGroups);
-				KWDataGridOptimizer::GetProfiler()->EndMethod("Post-optimization IV");
+				NEWKWDataGridOptimizer::GetProfiler()->EndMethod("Post-optimization IV");
 
 				if (bImprovement)
 				{
@@ -1856,7 +1856,8 @@ double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid*
 							     << endl;
 							neighbourDataGrid->Write(cout);
 						}
-						dCost = dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid);
+						dNeighbourDataGridCost =
+						    dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid);
 					}
 
 					// On remplace mergedDataGrid par la nouvelle grille integrant les
@@ -1882,7 +1883,7 @@ double KWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(const KWDataGrid*
 				cout << "Fin PostOptimisation VarPart" << endl;
 		}
 	}
-	ensure(fabs(dCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
+	ensure(fabs(dNeighbourDataGridCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
 	return dMergedCost;
 }
 

diff --git a/src/Learning/KWDataPreparation/KWDataGridOptimizer.h b/src/Learning/KWDataPreparation/KWDataGridOptimizer.h
@@ -252,7 +252,10 @@ class KWDataGridVNSOptimizer : public Object
 	// - neighbourDataGrid: grille courante en cours d'optimisation
 	// - dNeighbourDataGridCost: cout de la grille courante
 	// - mergedDataGrid: grille optimisee si amelioration
-	// En sortie, on rend la nouvelle version de la grille optimisee (ou courante) integrant la post-ptimisation et on renvoie son cout
+	// En sortie, la grille courante et son cout sont modifies suite a optimisation.
+	// On rend la grille issue de la post-opimisation VarPart, ainsi que sa grille partitionned re drefrence
+	// Le code retour est le meilleurs cout optenu apres post-optimisation VarPart
+	// Attention, celui-ci est different de celui de la grille courante optimisee
 	//
 	// A terme, il faudra isoler ce service de post-optimisation pour le deplacer en quatrime sous-methode
 	// en fin de la methode OptimizeSolution
@@ -261,7 +264,7 @@ class KWDataGridVNSOptimizer : public Object
 	// - Post-optimization
 	// - Post-optimization IV
 	double VNSDataGridPostOptimizeVarPart(const KWDataGrid* initialDataGrid, KWDataGridMerger* neighbourDataGrid,
-					      double dNeighbourDataGridCost, KWDataGrid* mergedDataGrid,
+					      double& dNeighbourDataGridCost, KWDataGrid* mergedDataGrid,
 					      KWDataGrid* partitionedReferencePostMergedDataGrid) const;
 
 	// CH IV Begin

diff --git a/src/Learning/KWDataPreparation/NEWKWDataGridOptimizer.cpp b/src/Learning/KWDataPreparation/NEWKWDataGridOptimizer.cpp
@@ -1640,8 +1640,8 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeDataGrid(const KWDataGrid* initialD
 
 // CH IV Begin
 double NEWKWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(
-    const KWDataGrid* initialDataGrid, KWDataGridMerger* neighbourDataGrid, double& dCost, KWDataGrid* mergedDataGrid,
-    KWDataGrid* partitionedReferencePostMergedDataGrid) const
+    const KWDataGrid* initialDataGrid, KWDataGridMerger* neighbourDataGrid, double& dNeighbourDataGridCost,
+    KWDataGrid* mergedDataGrid, KWDataGrid* partitionedReferencePostMergedDataGrid) const
 {
 	double dMergedCost;
 	double dFusionDeltaCost;
@@ -1653,11 +1653,10 @@ double NEWKWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(
 	require(initialDataGrid != NULL);
 	require(neighbourDataGrid != NULL);
 	require(neighbourDataGrid->GetDataGridCosts() == dataGridCosts);
-	require(fabs(dCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
+	require(fabs(dNeighbourDataGridCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
 
-	// Initialisations
-	dataGridManager.SetSourceDataGrid(initialDataGrid);
-	dMergedCost = dCost;
+	// Initialisation du cout
+	dMergedCost = dNeighbourDataGridCost;
 
 	// CH AB AF a voir avec Marc
 	// Cas ou l'on peut fusionner les parties de variable des cluster
@@ -1670,28 +1669,30 @@ double NEWKWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(
 		// Tri des attributs
 		neighbourDataGrid->SortAttributeParts();
 
+		// Initialisation du datagrid manager
 		dataGridManager.SetSourceDataGrid(neighbourDataGrid);
+
 		// Creation d'une nouvelle grille avec nouvelle description des PV fusionnees
 		dFusionDeltaCost =
 		    dataGridManager.ExportDataGridWithVarPartMergeOptimization(mergedDataGrid, dataGridCosts);
 		assert(not mergedDataGrid->GetVarPartsShared());
 
 		// Calcul et verification du cout
-		dMergedCost = dCost + dFusionDeltaCost;
+		dMergedCost = dNeighbourDataGridCost + dFusionDeltaCost;
 
 		// Cas ou le cout de la grille avec PV voisines fusionnees est plus eleve que le cout avant
 		// fusion
-		if (dMergedCost > dCost * (1 + dEpsilon) and bDisplayResults)
+		if (dMergedCost > dNeighbourDataGridCost * (1 + dEpsilon) and bDisplayResults)
 		{
 			sLabel = "PROBLEME : degradation du cout lors de la fusion des parties de variables "
 				 "contigues";
 			sLabel += DoubleToString(dMergedCost);
 			sLabel += "\t";
-			sLabel += DoubleToString(dMergedCost - dCost);
+			sLabel += DoubleToString(dMergedCost - dNeighbourDataGridCost);
 			AddWarning(sLabel);
 			cout << "PROBLEME : degradation du cout lors de la fusion des parties de variables "
 				"contigues : cout fusionne\t "
-			     << dMergedCost << "\tcout\t" << dCost << "\n";
+			     << dMergedCost << "\tcout\t" << dNeighbourDataGridCost << "\n";
 			cout << "Grille avant fusion\n";
 			neighbourDataGrid->Write(cout);
 			cout << "Grille apres fusion\n";
@@ -1828,7 +1829,8 @@ double NEWKWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(
 							     << endl;
 							neighbourDataGrid->Write(cout);
 						}
-						dCost = dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid);
+						dNeighbourDataGridCost =
+						    dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid);
 					}
 
 					// On remplace mergedDataGrid par la nouvelle grille integrant les
@@ -1854,7 +1856,7 @@ double NEWKWDataGridVNSOptimizer::VNSDataGridPostOptimizeVarPart(
 				cout << "Fin PostOptimisation VarPart" << endl;
 		}
 	}
-	ensure(fabs(dCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
+	ensure(fabs(dNeighbourDataGridCost - dataGridCosts->ComputeDataGridTotalCost(neighbourDataGrid)) < dEpsilon);
 	return dMergedCost;
 }
 
@@ -1886,8 +1888,6 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 	// Test du remplacement de la methode actuelle, par son proto
 	boolean bDeprecatedVersion = false;
 	//DDD bDeprecatedVersion = true;
-	cout << "DEPRECATED: (G=" << initialDataGrid->GetGranularity()
-	     << ", T=" << initialDataGrid->GetInnerAttributes()->GetVarPartGranularity() << ")" << endl;
 	if (bDeprecatedVersion)
 		return DEPRECATED_VNSOptimizeVarPartDataGrid(initialDataGrid, dDecreaseFactor, nMinIndex, nMaxIndex,
 							     optimizedDataGrid, dOptimizedDataGridCost,
@@ -1925,9 +1925,6 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 		dCost = OptimizeSolution(initialDataGrid, &neighbourDataGrid);
 		NEWKWDataGridOptimizer::GetProfiler()->EndMethod("Optimize solution");
 
-		//DDD
-		cout << "dCost\t" << dCost << endl;
-
 		// Post-optimisation des parties de variables de la grille
 		// A terme, a deplacer dans OptimizeSolution
 		dMergedCost = dCost;
@@ -1936,21 +1933,12 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 			    VNSDataGridPostOptimizeVarPart(initialDataGrid, &neighbourDataGrid, dCost, &mergedDataGrid,
 							   &partitionedReferencePostMergedDataGrid);
 
-		//DDD
-		cout << dOptimizedDataGridCost << "\t" << dCost << "\t" << dMergedCost << "\t" << dBestMergedCost
-		     << "\t" << (dMergedCost < dBestMergedCost - dEpsilon) << endl;
-
 		// Si amelioration: on la memorise
 		if (dMergedCost < dBestMergedCost - dEpsilon)
 		{
 			dBestCost = dCost;
 			dBestMergedCost = dMergedCost;
 
-			//DDD
-			cout << "\tNEW BEST\t" << dBestCost << "\t" << dBestMergedCost << endl;
-			cout << "\tneighbourDataGrid(" << neighbourDataGrid.GetInformativeAttributeNumber() << ", "
-			     << dataGridCosts->ComputeDataGridTotalCost(&neighbourDataGrid) << ")" << endl;
-
 			// Sauvegarde de la meilleure solution
 			if (bCleanNonInformativeVariables)
 			{
@@ -1961,27 +1949,17 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 				// reference est necessaire pour HandleOptimizationStep
 				dataGridManager.CopyInformativeDataGrid(&neighbourDataGrid, optimizedDataGrid);
 			}
-
 			else
 			{
 				dataGridManager.CopyDataGrid(&neighbourDataGrid, optimizedDataGrid);
 			}
 
-			//DDD
-			cout << "\toptimizedDataGrid(" << optimizedDataGrid->GetInformativeAttributeNumber() << ", "
-			     << dataGridCosts->ComputeDataGridTotalCost(optimizedDataGrid) << ")" << endl;
-
 			// Gestion de la meilleure solution
 			if (dataGridOptimizer != NULL)
 			{
-				//DDD
-				cout << "\tmergedDataGrid(" << mergedDataGrid.GetInformativeAttributeNumber() << ", "
-				     << dataGridCosts->ComputeDataGridTotalCost(&mergedDataGrid) << ")" << endl;
-
 				if (mergedDataGrid.GetInformativeAttributeNumber() == 0)
 					dataGridOptimizer->HandleOptimizationStep(optimizedDataGrid, initialDataGrid,
 										  false);
-
 				else
 				{
 					dataGridManager.SetSourceDataGrid(
@@ -1991,22 +1969,6 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 					dataGridOptimizer->HandleOptimizationStep(
 					    &mergedDataGrid, &partitionedReferencePostMergedDataGrid, false);
 
-					//DDD
-					cout << "\tGetDataGridOptimizer()->GetInitialVarPartDataGrid()("
-					     << GetDataGridOptimizer()
-						    ->GetInitialVarPartDataGrid()
-						    ->GetInformativeAttributeNumber()
-					     << ", "
-					     << dataGridCosts->ComputeDataGridTotalCost(
-						    GetDataGridOptimizer()->GetInitialVarPartDataGrid())
-					     << ")" << endl;
-					cout << "\tpartitionedReferencePostMergedDataGrid("
-					     << partitionedReferencePostMergedDataGrid.GetInformativeAttributeNumber()
-					     << ", "
-					     << dataGridCosts->ComputeDataGridTotalCost(
-						    &partitionedReferencePostMergedDataGrid)
-					     << ")" << endl;
-
 					// Nettoyage
 					partitionedReferencePostMergedDataGrid.DeleteAll();
 				}
@@ -2032,10 +1994,6 @@ double NEWKWDataGridVNSOptimizer::VNSOptimizeVarPartDataGrid(const KWDataGrid* i
 			break;
 	}
 
-	//DDD
-	cout << "=>\t" << dBestMergedDataGridCost << "\t" << dBestMergedCost << "\t" << dBestCost << "\t"
-	     << dataGridCosts->ComputeDataGridTotalCost(optimizedDataGrid) << endl;
-
 	// Memorisation du meilleur cout parmi les grilles post-fusionnees
 	dBestMergedDataGridCost = dBestMergedCost;
 	ensure(fabs(dBestCost - dataGridCosts->ComputeDataGridTotalCost(optimizedDataGrid)) < dEpsilon);
@@ -2100,9 +2058,6 @@ double NEWKWDataGridVNSOptimizer::DEPRECATED_VNSOptimizeVarPartDataGrid(const KW
 		dCost = OptimizeSolution(initialDataGrid, &neighbourDataGrid);
 		NEWKWDataGridOptimizer::GetProfiler()->EndMethod("Optimize solution");
 
-		//DDD
-		cout << "dCost\t" << dCost << endl;
-
 		// CH AB AF a voir avec Marc
 		// Cas ou l'on peut fusionner les parties de variable des cluster
 		// Probleme rencontre pour fusionner une grille non informative
@@ -2309,21 +2264,12 @@ double NEWKWDataGridVNSOptimizer::DEPRECATED_VNSOptimizeVarPartDataGrid(const KW
 		else
 			dMergedCost = dCost;
 
-		//DDD
-		cout << dOptimizedDataGridCost << "\t" << dCost << "\t" << dMergedCost << "\t" << dBestMergedCost
-		     << "\t" << (dMergedCost < dBestMergedCost - dEpsilon) << endl;
-
 		// Si amelioration: on la memorise
 		if (dMergedCost < dBestMergedCost - dEpsilon)
 		{
 			dBestCost = dCost;
 			dBestMergedCost = dMergedCost;
 
-			//DDD
-			cout << "\tNEW BEST\t" << dBestCost << "\t" << dBestMergedCost << endl;
-			cout << "\tneighbourDataGrid(" << neighbourDataGrid.GetInformativeAttributeNumber() << ", "
-			     << dataGridCosts->ComputeDataGridTotalCost(&neighbourDataGrid) << ")" << endl;
-
 			// Sauvegarde de la meilleure solution
 			if (bCleanNonInformativeVariables)
 			{
@@ -2339,17 +2285,9 @@ double NEWKWDataGridVNSOptimizer::DEPRECATED_VNSOptimizeVarPartDataGrid(const KW
 				dataGridManager.CopyDataGrid(&neighbourDataGrid, optimizedDataGrid);
 			}
 
-			//DDD
-			cout << "\toptimizedDataGrid(" << optimizedDataGrid->GetInformativeAttributeNumber() << ", "
-			     << dataGridCosts->ComputeDataGridTotalCost(optimizedDataGrid) << ")" << endl;
-
 			// Gestion de la meilleure solution
 			if (dataGridOptimizer != NULL)
 			{
-				//DDD
-				cout << "\tmergedDataGrid(" << mergedDataGrid.GetInformativeAttributeNumber() << ", "
-				     << dataGridCosts->ComputeDataGridTotalCost(&mergedDataGrid) << ")" << endl;
-
 				if (mergedDataGrid.GetInformativeAttributeNumber() == 0)
 					dataGridOptimizer->HandleOptimizationStep(optimizedDataGrid, initialDataGrid,
 										  false);
@@ -2362,22 +2300,6 @@ double NEWKWDataGridVNSOptimizer::DEPRECATED_VNSOptimizeVarPartDataGrid(const KW
 					dataGridOptimizer->HandleOptimizationStep(
 					    &mergedDataGrid, &partitionedReferencePostMergedDataGrid, false);
 
-					//DDD
-					cout << "\tGetDataGridOptimizer()->GetInitialVarPartDataGrid()("
-					     << GetDataGridOptimizer()
-						    ->GetInitialVarPartDataGrid()
-						    ->GetInformativeAttributeNumber()
-					     << ", "
-					     << dataGridCosts->ComputeDataGridTotalCost(
-						    GetDataGridOptimizer()->GetInitialVarPartDataGrid())
-					     << ")" << endl;
-					cout << "\tpartitionedReferencePostMergedDataGrid("
-					     << partitionedReferencePostMergedDataGrid.GetInformativeAttributeNumber()
-					     << ", "
-					     << dataGridCosts->ComputeDataGridTotalCost(
-						    &partitionedReferencePostMergedDataGrid)
-					     << ")" << endl;
-
 					// Nettoyage
 					partitionedReferencePostMergedDataGrid.DeleteAll();
 				}
@@ -2404,10 +2326,6 @@ double NEWKWDataGridVNSOptimizer::DEPRECATED_VNSOptimizeVarPartDataGrid(const KW
 			break;
 	}
 
-	//DDD
-	cout << "=>\t" << dBestMergedDataGridCost << "\t" << dBestMergedCost << "\t" << dBestCost << "\t"
-	     << dataGridCosts->ComputeDataGridTotalCost(optimizedDataGrid) << endl;
-
 	// Memorisation du meilleur cout parmi les grilles post-fusionnees
 	dBestMergedDataGridCost = dBestMergedCost;