# labels = data[:, -1]
# print("dataset = %s data.shape = %s" % (dataset, fea.shape))
fea, labels = loadData.load_coil100()
print("------ Normalizing data ------")
# fea = tool.data_Normalized(fea)
Normalizer = MinMaxScaler()
Normalizer.fit(fea)
fea = Normalizer.transform(fea)
print("------ PCA decomposition ------")
# fea,b,c = tool.PCA.pca(fea, 150)
pca = PCA(n_components=150)
fea = pca.fit_transform(fea)
print("fea.shape =", fea.shape)
K = 5
groupNumber = len(np.unique(labels))
print("------ Clustering ------")
start = time.time()
cl = knnDPC.knnDPC2(fea, groupNumber, K)
end = time.time()
print("time =", end - start)
print("------ Computing performance measure ------")
nmi = measure.NMI(labels, cl)
print("nmi =", nmi)
acc = measure.ACC(labels, cl) # 计算ACC需要label和labels_pred 在同一个区间[1,20]
print("acc =", acc)
print("data_set = %s data.shape = %s" % (data_set, fea.shape))
print("------ Normalizing data ------")
# tool.data_Normalized(fea)
Normalizer = MinMaxScaler()
Normalizer.fit(fea)
fea = Normalizer.transform(fea)
print("------ Decomposition ------")
#fea,b,c = PCA.pca(fea, 150)
pca = PCA(n_components=150)
fea = pca.fit_transform(fea)
print("fea.shape =", fea.shape)
K = 20
u = 1
group_number = len(np.unique(labels))
print("------ Clustering ------")
start = time.time()
labels_pred = DPC.knnDPC1(fea, group_number, K, u)
end = time.time()
print("time =", end - start)
nmi = measure.NMI(labels, labels_pred)
print("nmi =", nmi)
acc = measure.ACC(labels, labels_pred)
print("acc =", acc)
precision_score = measure.precision_score(labels, labels_pred)
print("precision_score =", precision_score)
K = 10
v = 1
z = 0.01
groupNumber = len(np.unique(labels))
ND = dist.shape[0]
# build the adjacency graph
graphW, NNIndex = tool.gacBuildDigraph(dist, K, v)
graphW = np.around(graphW, decimals=4)
# from adjacency matrix to probability transition matrix
def f(x):
return x / np.sum(x)
graphW = np.apply_along_axis(f, 1, graphW)
initialCluster = tool.gacNNMerge(dist, NNIndex)
numClusters = len(initialCluster)
cl = tool.gacMerging(graphW, initialCluster, groupNumber, 'path', z)
end = time.time()
print("time =", end - start)
print("------ Computing performance measure ------")
NMI = measure.NMI(labels, cl)
print("NMI =", NMI)
ACC = measure.ACC(labels, cl)
print("ACC =", ACC)
