def hdbscan():
global source_data
min_cluster_size = 0.5
min_samples = 5
alpha = 1.0
cluster_selection_method = 'eom'
if request.args.get('min_cluster_size'):
min_cluster_size = int(request.args.get('min_cluster_size'))
if request.args.get('min_samples'):
min_samples = int(request.args.get('min_samples'))
if request.args.get('alpha'):
alpha = float(request.args.get('alpha'))
if request.args.get('cluster_selection_method'):
cluster_selection_method = int(
request.args.get('cluster_selection_method'))
model = MyHDBSCAN()
clustering = model.fit(source_data, min_cluster_size, min_samples, alpha,
cluster_selection_method)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='HDBSCAN')
def pca():
model = MyPCA(source_data)
projection_data = model.return_data
projection_data = projection_data.tolist()
return render_template("pca.html",
data=projection_data,
data_obj=dictionary_data,
method='PCA')
def hierarchical():
method = 'average'
if request.args.get('linkage'):
method = request.args.get('linkage')
model = MyHierarchical()
clustering, labels = model.fit(source_data, method)
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='hierarchy')
def sc():
n_clusters = 8
if request.args.get('s_clusters'):
n_clusters = int(request.args.get('s_clusters'))
model = MySC()
clustering = model.fit(source_data, n_clusters)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='SpectralClustering')
def meanshift():
global source_data
bandwidth = None
if request.args.get('bandwidth'):
bandwidth = float(request.args.get('bandwidth'))
model = MyMeanShift()
clustering = model.fit(source_data, bandwidth)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='MeanShift')
def gmm():
n_clusters = 1
cov_type = 'full'
if request.args.get('g_clusters'):
n_clusters = int(request.args.get('g_clusters'))
if request.args.get('cov_type'):
cov_type = request.args.get('cov_type')
model = MyGMM()
clustering, labels = model.fit(source_data, n_clusters, cov_type)
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='GaussianMixture')
def fa():
n_clusters = 2
method = 'ward'
if request.args.get('f_clusters'):
n_clusters = int(request.args.get('f_clusters'))
if request.args.get('linkage'):
method = request.args.get('linkage')
model = MyFeatureAgglomeration()
clustering = model.fit(source_data, n_clusters, method)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='FeatureAgglomeration')
def dbscan():
eps = 0.5
min_samples = 5
if request.args.get('eps'):
eps = float(request.args.get('eps'))
if request.args.get('min_samples'):
min_samples = int(request.args.get('min_samples'))
model = MyDBSCAN()
clustering = model.fit(source_data, eps, min_samples)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='DBSCAN')
def birch():
threshold = 0.5
branching_factor = 50
if request.args.get('threshold'):
threshold = float(request.args.get('threshold'))
if request.args.get('branching_factor'):
branching_factor = int(request.args.get('branching_factor'))
model = MyBirch()
clustering = model.fit(source_data, threshold, branching_factor)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='Birch')
def ap():
damping = 0.5
preference = None
if request.args.get('damping'):
damping = float(request.args.get('damping'))
if request.args.get('preference'):
preference = float(request.args.get('preference'))
model = MyAP()
clustering = model.fit(source_data, damping, preference)
labels = clustering.labels_
labels = labels.tolist()
clusters = np.unique(labels).size
pca = MyPCA(source_data)
data_pca = pca.return_data
samples, features = data_pca.shape
data_pca = data_pca.tolist()
for i in range(samples):
data_pca[i].append(labels[i])
return render_template("cluster.html",
data=data_pca,
data_obj=dictionary_data,
clusters=clusters,
method='AffinityPropagation')