def cartesBienClusterisees(couche="conv1", seuilSuppression=559, seuilBonClustering=30, fichier=False): pFRJA_R_KMNI, pFRJA_V_KMNI, pFR_RV_KMNI, pCIC_R_KMNI, pCIC_V_KMNI, ind = MapsClustering( couche, seuilSuppression, "kmeansNonInit", fichier) pFRJA_R_KMI, pFRJA_V_KMI, pFR_RV_KMI, pCIC_R_KMI, pCIC_V_KMI, indKmeansInit = MapsClustering( couche, seuilSuppression, "kmeansInit", fichier) pFRJA_R_DBSCAN, pFRJA_V_DBSCAN, pFR_RV_DBSCAN, pCIC_R_DBSCAN, pCIC_V_DBSCAN, indDBSCAN = MapsClustering( couche, seuilSuppression, "DBSCAN", fichier) # pFRJA_R_MeanShift, pFRJA_V_MeanShift, pFR_RV_MeanShift, pCIC_R_MeanShift, pCIC_V_MeanShift, indMeanShift = MapsClustering(couche, seuilSuppression, "MeanShift", fichier) ################################################################################ #Initialisation des matrices ################################################################################ bienClusterisesKmeansNonInit = [] bienClusterisesKmeansInit = [] bienClusterisesDBSCAN = [] # bienClusterisesMeanShift = [] ################################################################################ #Clustering 1 ################################################################################ clus = bienClusterise(MatriceClustering=pFRJA_R_KMNI, seuil=seuilBonClustering, indices=ind[0]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[0]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[0]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[0]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 1bis ################################################################################ clus = bienClusterise(MatriceClustering=pFRJA_V_KMNI, seuil=seuilBonClustering, indices=ind[1]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[1]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[1]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[1]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 2 ################################################################################ clus = bienClusterise(MatriceClustering=pFR_RV_KMNI, seuil=seuilBonClustering, indices=ind[2]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFR_RV_KMI, seuil=seuilBonClustering, indices=indKmeansInit[2]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFR_RV_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[2]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[2]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 3 ################################################################################ clus = bienClusterise(MatriceClustering=pCIC_R_KMNI, seuil=seuilBonClustering, indices=ind[3]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[3]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[3]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[3]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 3bis ################################################################################ clus = bienClusterise(MatriceClustering=pCIC_V_KMNI, seuil=seuilBonClustering, indices=ind[4]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[4]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[4]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[4]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Enregistrement des differentes matrices ################################################################################ np.save( '../resultats/' + couche + '/kmeansNonInit/bienClusterisekmeansNonInit.npy', bienClusterisesKmeansNonInit) np.save( '../resultats/' + couche + '/kmeansInit/bienClusterisekmeansInit.npy', bienClusterisesKmeansInit) np.save('../resultats/' + couche + '/DBSCAN/bienClusteriseDBSCAN.npy', bienClusterisesDBSCAN)
def GenerationClustering(couche="conv1", seuilSuppression=559, seuilBonClustering=30, fichier=True): ############################################################################################# # Appel de MapsClustering et enregistrement des matrices des indices pour chaque algorithme de chaque couche; # Matrice des indices : 5 colonnes representant les 5 clusterings, et dans chaque colonne les indices de cartes # qui donnent bien deux classes. ############################################################################################# pFRJA_R_KMNI, pFRJA_V_KMNI, pFR_RV_KMNI, pCIC_R_KMNI, pCIC_V_KMNI, ind = MapsClustering( couche, seuilSuppression, "kmeansNonInit", fichier) matKmeansNonInit = str( "../resultats/" + couche + "/kmeansNonInit/matKmeansNonInit_indices_bonnes_cartes.csv") f1 = open(matKmeansNonInit, "wb") writer = csv.writer(f1) writer.writerow([ "KmeansNonInit(1)", "KmeansNonInit(1bis)", "KmeansNonInit(2)", "KmeansNonInit(2bis)", "KmeansNonInit(3)" ]) for values in izip_longest(*ind): writer.writerow(values) pFRJA_R_KMI, pFRJA_V_KMI, pFR_RV_KMI, pCIC_R_KMI, pCIC_V_KMI, indKmeansInit = MapsClustering( couche, seuilSuppression, "kmeansInit", fichier) matKmeansInit = str("../resultats/" + couche + "/kmeansInit/matKmeansInit_indices_bonnes_cartes.csv") f2 = open(matKmeansInit, "wb") writer = csv.writer(f2) writer.writerow([ "KmeansInit(1)", "KmeansInit(1bis)", "KmeansInit(2)", "KmeansInit(2bis)", "KmeansInit(3)" ]) for values in izip_longest(*indKmeansInit): writer.writerow(values) pFRJA_R_DBSCAN, pFRJA_V_DBSCAN, pFR_RV_DBSCAN, pCIC_R_DBSCAN, pCIC_V_DBSCAN, indDBSCAN = MapsClustering( couche, seuilSuppression, "DBSCAN", fichier) matDBSCAN = str("../resultats/" + couche + "/DBSCAN/matDBSCAN_indices_bonnes_cartes.csv") f3 = open(matDBSCAN, "wb") writer = csv.writer(f3) writer.writerow([ "DBSCAN(1)", "DBSCAN(1bis)", "DBSCAN(2)", "DBSCAN(2bis)", "DBSCAN(3)" ]) for values in izip_longest(*indDBSCAN): writer.writerow(values) # pFRJA_R_MeanShift, pFRJA_V_MeanShift, pFR_RV_MeanShift, pCIC_R_MeanShift, pCIC_V_MeanShift, indMeanShift = MapsClustering(couche, seuilSuppression, "MeanShift", fichier) # matMeanShift = str("../resultats/" + couche + "/MeanShift/matMeanShift_indices_bonnes_cartes.csv") # f4 = open(matMeanShift, "wb") # writer = csv.writer(f4) # writer.writerow(["MeanShift(1)", "MeanShift(1bis)", "MeanShift(2)", "MeanShift(2bis)", "MeanShift(3)"]) # for values in izip_longest(*indMeanShift): # writer.writerow(values) ############################################################################################# # Fichiers de cartes bon clustering ############################################################################################# filename = str("../resultats/" + couche + "/cartes_bon_clustering22") f = open(filename, "wb") ############################################################################# #appel directement avec les matrices ############################################################################## f.write("FRJA R\n") clus = bienClusterise(MatriceClustering=pFRJA_R_KMNI, seuil=seuilBonClustering, indices=ind[0]) f.write("kmeansNonInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFRJA_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[0]) f.write("kmeansInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFRJA_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[0]) f.write("DBSCAN:" + str(clus) + "\n") # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering,indices= indMeanShift[0]) # f.write("MeanShift:" + str(clus)+"\n") f.write("FRJA V\n") clus = bienClusterise(MatriceClustering=pFRJA_V_KMNI, seuil=seuilBonClustering, indices=ind[1]) f.write("kmeansNonInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFRJA_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[1]) f.write("kmeansInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFRJA_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[1]) f.write("DBSCAN:" + str(clus) + "\n") # clus = bienClusterise(MatriceClustering=pFRJA_V_MeanShift, seuil=seuilBonClustering,indices= indMeanShift[1]) # f.write("MeanShift:" + str(clus)+"\n") f.write("FR RV\n") clus = bienClusterise(MatriceClustering=pFR_RV_KMNI, seuil=seuilBonClustering, indices=ind[2]) f.write("kmeansNonInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFR_RV_KMI, seuil=seuilBonClustering, indices=indKmeansInit[2]) f.write("kmeansInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pFR_RV_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[2]) f.write("DBSCAN:" + str(clus) + "\n") # clus = bienClusterise(MatriceClustering=pFR_RV_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[2]) # f.write("MeanShift:" + str(clus)+"\n") f.write("JA correct/incorrect R\n") clus = bienClusterise(MatriceClustering=pCIC_R_KMNI, seuil=seuilBonClustering, indices=ind[3]) f.write("kmeansNonInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pCIC_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[3]) f.write("kmeansInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pCIC_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[3]) f.write("DBSCAN:" + str(clus) + "\n") # clus = bienClusterise(MatriceClustering=pCIC_R_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[3]) # f.write("MeanShift:" + str(clus)+"\n") f.write("JA correct/incorrect V\n") clus = bienClusterise(MatriceClustering=pCIC_V_KMNI, seuil=seuilBonClustering, indices=ind[4]) f.write("kmeansNonInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pCIC_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[4]) f.write("kmeansInit:" + str(clus) + "\n") clus = bienClusterise(MatriceClustering=pCIC_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[4]) f.write("DBSCAN:" + str(clus) + "\n") # clus = bienClusterise(MatriceClustering=pCIC_V_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[4]) # f.write("MeanShift:" + str(clus)+"\n") f.close()
def GenerationClustering(couche = "conv1", seuilSuppression = 559, seuilBonClustering = 30, fichier = True): ############################################################################################# # Appel de MapsClustering et enregistrement des matrices des indices pour chaque algorithme de chaque couche; # Matrice des indices : 5 colonnes representant les 5 clusterings, et dans chaque colonne les indices de cartes # qui donnent bien deux classes. ############################################################################################# pFRJA_R_KMNI, pFRJA_V_KMNI, pFR_RV_KMNI, pCIC_R_KMNI, pCIC_V_KMNI, ind = MapsClustering(couche, seuilSuppression, "kmeansNonInit", fichier) matKmeansNonInit = str("../resultats/" + couche + "/kmeansNonInit/matKmeansNonInit_indices_bonnes_cartes.csv") f1 = open(matKmeansNonInit, "wb") writer = csv.writer(f1) writer.writerow(["KmeansNonInit(1)", "KmeansNonInit(1bis)", "KmeansNonInit(2)", "KmeansNonInit(2bis)", "KmeansNonInit(3)"]) for values in izip_longest(*ind): writer.writerow(values) pFRJA_R_KMI, pFRJA_V_KMI, pFR_RV_KMI, pCIC_R_KMI, pCIC_V_KMI, indKmeansInit = MapsClustering(couche, seuilSuppression, "kmeansInit", fichier) matKmeansInit = str("../resultats/" + couche + "/kmeansInit/matKmeansInit_indices_bonnes_cartes.csv") f2 = open(matKmeansInit, "wb") writer = csv.writer(f2) writer.writerow(["KmeansInit(1)", "KmeansInit(1bis)", "KmeansInit(2)", "KmeansInit(2bis)", "KmeansInit(3)"]) for values in izip_longest(*indKmeansInit): writer.writerow(values) pFRJA_R_DBSCAN, pFRJA_V_DBSCAN, pFR_RV_DBSCAN, pCIC_R_DBSCAN, pCIC_V_DBSCAN, indDBSCAN = MapsClustering(couche, seuilSuppression, "DBSCAN", fichier) matDBSCAN = str("../resultats/" + couche + "/DBSCAN/matDBSCAN_indices_bonnes_cartes.csv") f3 = open(matDBSCAN, "wb") writer = csv.writer(f3) writer.writerow(["DBSCAN(1)", "DBSCAN(1bis)", "DBSCAN(2)", "DBSCAN(2bis)", "DBSCAN(3)"]) for values in izip_longest(*indDBSCAN): writer.writerow(values) # pFRJA_R_MeanShift, pFRJA_V_MeanShift, pFR_RV_MeanShift, pCIC_R_MeanShift, pCIC_V_MeanShift, indMeanShift = MapsClustering(couche, seuilSuppression, "MeanShift", fichier) # matMeanShift = str("../resultats/" + couche + "/MeanShift/matMeanShift_indices_bonnes_cartes.csv") # f4 = open(matMeanShift, "wb") # writer = csv.writer(f4) # writer.writerow(["MeanShift(1)", "MeanShift(1bis)", "MeanShift(2)", "MeanShift(2bis)", "MeanShift(3)"]) # for values in izip_longest(*indMeanShift): # writer.writerow(values) ############################################################################################# # Fichiers de cartes bon clustering ############################################################################################# filename = str("../resultats/" + couche + "/cartes_bon_clustering_MeanShift") f = open(filename, "wb") ############################################################################# #appel directement avec les matrices ############################################################################## f.write("FRJA R\n") clus = bienClusterise(MatriceClustering=pFRJA_R_KMNI, seuil=seuilBonClustering, indices=ind[0]) f.write("kmeansNonInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFRJA_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[0]) f.write("kmeansInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFRJA_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[0]) f.write("DBSCAN:"+ str(clus)+"\n") # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering,indices= indMeanShift[0]) # f.write("MeanShift:" + str(clus)+"\n") f.write("FRJA V\n") clus = bienClusterise(MatriceClustering=pFRJA_V_KMNI, seuil=seuilBonClustering, indices=ind[1]) f.write("kmeansNonInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFRJA_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[1]) f.write("kmeansInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFRJA_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[1]) f.write("DBSCAN:" + str(clus)+"\n") # clus = bienClusterise(MatriceClustering=pFRJA_V_MeanShift, seuil=seuilBonClustering,indices= indMeanShift[1]) # f.write("MeanShift:" + str(clus)+"\n") f.write("FR RV\n") # clus = bienClusterise(MatriceClustering=pFR_RV_KMNI, seuil=seuilBonClustering, indices=ind[2]) f.write("kmeansNonInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFR_RV_KMI, seuil=seuilBonClustering,indices=indKmeansInit[2]) f.write("kmeansInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pFR_RV_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[2]) f.write("DBSCAN:" + str(clus)+"\n") # clus = bienClusterise(MatriceClustering=pFR_RV_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[2]) # f.write("MeanShift:" + str(clus)+"\n") f.write("JA correct/incorrect R\n") clus = bienClusterise(MatriceClustering=pCIC_R_KMNI, seuil=seuilBonClustering, indices=ind[3]) f.write("kmeansNonInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pCIC_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[3]) f.write("kmeansInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pCIC_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[3]) f.write("DBSCAN:"+ str(clus)+"\n") # clus = bienClusterise(MatriceClustering=pCIC_R_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[3]) # f.write("MeanShift:" + str(clus)+"\n") f.write("JA correct/incorrect V\n") clus = bienClusterise(MatriceClustering=pCIC_V_KMNI, seuil=seuilBonClustering, indices=ind[4]) f.write("kmeansNonInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pCIC_V_KMI, seuil=seuilBonClustering,indices=indKmeansInit[4]) f.write("kmeansInit:" + str(clus)+"\n") clus = bienClusterise(MatriceClustering=pCIC_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[4]) f.write("DBSCAN:" + str(clus)+"\n") # clus = bienClusterise(MatriceClustering=pCIC_V_MeanShift, seuil=seuilBonClustering, indices= indMeanShift[4]) # f.write("MeanShift:" + str(clus)+"\n") f.close()
def cartesBienClusterisantes(couche = "conv1", seuilSuppression = 559, seuilBonClustering = 30, fichier = False): pFRJA_R_KMNI, pFRJA_V_KMNI, pFR_RV_KMNI, pCIC_R_KMNI, pCIC_V_KMNI, ind = MapsClustering(couche, seuilSuppression, "kmeansNonInit", fichier) pFRJA_R_KMI, pFRJA_V_KMI, pFR_RV_KMI, pCIC_R_KMI, pCIC_V_KMI, indKmeansInit = MapsClustering(couche, seuilSuppression, "kmeansInit", fichier) pFRJA_R_DBSCAN, pFRJA_V_DBSCAN, pFR_RV_DBSCAN, pCIC_R_DBSCAN, pCIC_V_DBSCAN, indDBSCAN = MapsClustering(couche, seuilSuppression, "DBSCAN", fichier) # pFRJA_R_MeanShift, pFRJA_V_MeanShift, pFR_RV_MeanShift, pCIC_R_MeanShift, pCIC_V_MeanShift, indMeanShift = MapsClustering(couche, seuilSuppression, "MeanShift", fichier) ################################################################################ #Initialisation des matrices ################################################################################ bienClusterisesKmeansNonInit = [] bienClusterisesKmeansInit = [] bienClusterisesDBSCAN = [] # bienClusterisesMeanShift = [] ################################################################################ #Clustering 1 ################################################################################ clus = bienClusterise(MatriceClustering=pFRJA_R_KMNI, seuil=seuilBonClustering, indices=ind[0]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[0]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[0]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[0]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 1bis ################################################################################ clus = bienClusterise(MatriceClustering=pFRJA_V_KMNI, seuil=seuilBonClustering, indices=ind[1]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_V_KMI, seuil=seuilBonClustering, indices=indKmeansInit[1]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFRJA_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[1]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[1]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 2 ################################################################################ clus = bienClusterise(MatriceClustering=pFR_RV_KMNI, seuil=seuilBonClustering, indices=ind[2]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pFR_RV_KMI, seuil=seuilBonClustering,indices=indKmeansInit[2]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pFR_RV_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[2]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[2]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 3 ################################################################################ clus = bienClusterise(MatriceClustering=pCIC_R_KMNI, seuil=seuilBonClustering, indices=ind[3]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_R_KMI, seuil=seuilBonClustering, indices=indKmeansInit[3]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_R_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[3]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[3]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Clustering 3bis ################################################################################ clus = bienClusterise(MatriceClustering=pCIC_V_KMNI, seuil=seuilBonClustering, indices=ind[4]) bienClusterisesKmeansNonInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_V_KMI, seuil=seuilBonClustering,indices=indKmeansInit[4]) bienClusterisesKmeansInit.append(clus) clus = bienClusterise(MatriceClustering=pCIC_V_DBSCAN, seuil=seuilBonClustering, indices=indDBSCAN[4]) bienClusterisesDBSCAN.append(clus) # clus = bienClusterise(MatriceClustering=pFRJA_R_MeanShift, seuil=seuilBonClustering, indices=indMeanShift[4]) # bienClusterisesMeanShift.append(clus) ################################################################################ #Enregistrement des differentes matrices ################################################################################ np.save('../resultats/' + couche + '/kmeansNonInit/bienClusteriseKmeansNonInit.npy', bienClusterisesKmeansNonInit) np.save('../resultats/' + couche + '/kmeansInit/bienClusteriseKmeansInit.npy', bienClusterisesKmeansInit) np.save('../resultats/' + couche + '/DBSCAN/bienClusteriseDBSCAN.npy', bienClusterisesDBSCAN) # np.save('../resultats/' + couche + '/MeanShift/bienClusteriseMeanShift.npy', bienClusterisesMeanShift)
def GenerationClustering(couche = "conv1", seuilSuppression = 559, seuilBonClustering = 30, fichier = True): ############################################################################################# # Appel de MapsClustering et enregistrement des matrices des indices pour chaque algorithme de chaque couche; # Matrice des indices : 5 colonnes representant les 5 clusterings, et dans chaque colonne les indices de cartes # qui donnent bien deux classes. ############################################################################################# pFRJA_R_KMNI, pFRJA_V_KMNI, pFR_RV_KMNI, pCIC_R_KMNI, pCIC_V_KMNI, pourcentagesFR_Rlvb0, pourcentagesFR_Rlvb1,pourcentagesFR_Rlvb2, pourcentagesFR_Rlvb3, ind = Maps5Clustering(couche, seuilSuppression, fichier) matKmeansNonInit = str("../resultats/resultats5phonemes/matKmeansNonInit_indices_bonnes_cartes.csv") f1 = open(matKmeansNonInit, "wb") writer = csv.writer(f1) writer.writerow(["KmeansNonInit(1)", "KmeansNonInit(1bis)", "KmeansNonInit(2)", "KmeansNonInit(3)", "KmeansNonInit(3bis)", " 4phonemes"]) for values in izip_longest(*ind): writer.writerow(values) ############################################################################################# # Fichiers de cartes bon clustering ############################################################################################# filename = str("../resultats/resultats5phonemes/cartes_bon_clustering_kmeans5") f = open(filename, "wb") ############################################################################# #appel directement avec les matrices ############################################################################## f.write("FRJA R\n") clus = bienClusterise(MatriceClustering=pFRJA_R_KMNI, seuil=seuilBonClustering, indices=ind[0]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("FRJA V\n") clus = bienClusterise(MatriceClustering=pFRJA_V_KMNI, seuil=seuilBonClustering, indices=ind[1]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("FR RV\n") clus = bienClusterise(MatriceClustering=pFR_RV_KMNI, seuil=seuilBonClustering, indices=ind[2]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("JA correct/incorrect R\n") # clus = bienClusterise(MatriceClustering=pCIC_R_KMNI, seuil=seuilBonClustering, indices=ind[3]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("JA correct/incorrect V\n") # clus = bienClusterise(MatriceClustering=pCIC_V_KMNI, seuil=seuilBonClustering, indices=ind[4]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("FR_Rlvb\n") f.write("classe0") clus = bienClusterise(MatriceClustering=pourcentagesFR_Rlvb0, seuil=seuilBonClustering, indices=ind[5]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("classe1") clus = bienClusterise(MatriceClustering=pourcentagesFR_Rlvb1, seuil=seuilBonClustering, indices=ind[5]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("classe2") clus = bienClusterise(MatriceClustering=pourcentagesFR_Rlvb2, seuil=seuilBonClustering, indices=ind[5]) f.write("kmeansNonInit:" + str(clus)+"\n") f.write("classe3") clus = bienClusterise(MatriceClustering=pourcentagesFR_Rlvb3, seuil=seuilBonClustering, indices=ind[5]) f.write("kmeansNonInit:" + str(clus)+"\n") f.close()