def create_kmedoid_clustering(doc2vec_model_file_path, tsne_model_file_path, start_medoids, new_hostnames=None):
"""helper function to create K-Medoid clustering plot
:param doc2vec_model_file_path: file path of doc2vec model
:param tsne_model_file_path: file path of tsne model
other parameters are explained in dbscan_clustering function
"""
doc2vec_model = doc2vec.load_existing_model(doc2vec_model_file_path=doc2vec_model_file_path)
tsne_model = tsne.load_tsne_model(tsne_model_file_path=tsne_model_file_path)
kmedoid_clustering(doc2vec_model, tsne_model, start_medoids, new_hostnames=new_hostnames)
def create_affinity_propagation_cluster(doc2vec_model_file_path, tsne_model_file_path, model_language,
save_to_directory, new_hostnames=None):
"""helper function to create affinity propagation clustering plot
:param doc2vec_model_file_path: file path of doc2vec model
:param tsne_model_file_path: file path of tsne model
other parameters are explained in affinity_propagation_cluster function
"""
doc2vec_model = doc2vec.load_existing_model(doc2vec_model_file_path=doc2vec_model_file_path)
tsne_model = tsne.load_tsne_model(tsne_model_file_path=tsne_model_file_path)
affinity_propagation_cluster(doc2vec_model, tsne_model, model_language, new_hostnames, save_to_directory)
def main():
# example usage for create K-Medoid Clustering
doc2vec_model = doc2vec.load_existing_model(
model_file_name='doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t_sne_single_language_full_model_doc2vec_model_german_20_Feb_2018_08_56_12.npy')
kmedoid_clustering(doc2vec_model,
tsne_model,
start_medoids=[0, 5, 10, 15, 20, 25, 30, 35])
'''
def main():
# example usage for creating an affinity propagation cluster
doc2vec_model = doc2vec.load_existing_model(
model_file_name='doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t_sne_single_language_full_model_doc2vec_model_german_20_Feb_2018_08_56_12.npy')
affinity_propagation_cluster(doc2vec_model=doc2vec_model,
tsne_model=tsne_model,
model_language='single_language_full_german')
'''
def main():
# example usage for create Agglomerative Clustering
doc2vec_model = doc2vec.load_existing_model(
model_file_name=
'doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name=
't_sne_single_language_full_model_doc2vec_model_german_20_Feb_2018_08_56_12.npy'
)
agglomerative_clustering(
doc2vec_model,
tsne_model,
numbers_of_clusters=[5, 10, 15, 20, 25, 30],
model_language='single-language_full_model-german')
'''
def create_kmeans_clustering(doc2vec_model_file_path,
tsne_model_file_path,
model_language,
k,
save_to_directory,
new_hostnames=None):
"""helper function to create K-Means clustering plot
:param doc2vec_model_file_path: file path of doc2vec model
:param tsne_model_file_path: file path of tsne model
other parameters are explained in dbscan_clustering function
"""
doc2vec_model = doc2vec.load_existing_model(
doc2vec_model_file_path=doc2vec_model_file_path)
tsne_model = tsne.load_tsne_model(
tsne_model_file_path=tsne_model_file_path)
kmeans_clustering(doc2vec_model=doc2vec_model,
tsne_model=tsne_model,
model_language=model_language,
k=k,
new_hostnames=new_hostnames,
save_to_directory=save_to_directory)
def create_agglomerative_clustering(doc2vec_model_file_path,
tsne_model_file_path,
numbers_of_clusters,
model_language,
save_to_directory,
new_hostnames=None):
"""helper function to create agglomerative clustering plot
:param doc2vec_model_file_path: file path of doc2vec model
:param tsne_model_file_path: file path of tsne model
other parameters are explained in agglomerative_clustering function
"""
doc2vec_model = doc2vec.load_existing_model(
doc2vec_model_file_path=doc2vec_model_file_path)
tsne_model = tsne.load_tsne_model(
tsne_model_file_path=tsne_model_file_path)
agglomerative_clustering(doc2vec_model=doc2vec_model,
tsne_model=tsne_model,
numbers_of_clusters=numbers_of_clusters,
model_language=model_language,
new_hostnames=new_hostnames,
save_to_directory=save_to_directory)
def main():
'''
# example usage for create K-Means Clustering
doc2vec_model = doc2vec.load_existing_model(
model_file_name='doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t_sne_single_language_full_model_doc2vec_model_german_20_Feb_2018_08_56_12.npy')
kmeans_clustering(doc2vec_model,
tsne_model,
model_language='single_language_full_german',
k=10)
# example with K-Means and new added data points
doc2vec_model = doc2vec.load_existing_model(
model_file_name='doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t_sne_s_l_full_doc2vec_model_new_data_german_20_Feb_2018_09_58_50.npy')
kmeans_clustering(doc2vec_model,
tsne_model,
model_language='s-l-full-model-new-data-german',
new_hostnames=['upkbs.ch',
'curaneo.ch',
'bscyb.ch',
'scltigers.ch',
'graubuenden.ch'],
k=10)
'''
# Live-Demo pt. 1
doc2vec_model = doc2vec.load_existing_model(
model_file_name=
'doc2vec_single_language_full_model_german_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t-sne-live-demo-german-07-Mar-2018-09:08:12.npy')
kmeans_clustering(doc2vec_model,
tsne_model,
model_language='live-demo-german',
new_hostnames=['triaplus.ch', 'fcaarau.ch'],
k=10)
# Live-Demo pt. 2
doc2vec_model = doc2vec.load_existing_model(
model_file_name=
'doc2vec_single_language_full_model_english_18_Feb_2018_22_31_27')
tsne_model = tsne.load_tsne_model(
model_file_name='t-sne-live-demo-english-07-Mar-2018-09:08:18.npy')
kmeans_clustering(doc2vec_model,
tsne_model,
model_language='live-demo-english',
new_hostnames=['hostelscentral.com'],
k=10)