def get_cluster_quality():
"""Returns cluster quality.
"""
print('Getting vocabulary ...')
data_file = os.path.join(args.data_path, 'min_df_{}'.format(args.min_df))
vocab, cluster_valid = data.get_all_data(data_file, temporal=True)
vocab_size = len(vocab)
topics_distributions = []
# get data
print('Getting full data ...')
tokens = train['tokens']
counts = train['counts']
times = train['times']
num_times = len(np.unique(train_times))
num_docs = len(tokens)
rnn_inp = data.get_rnn_input(tokens, counts, times, num_times, vocab_size, num_docs)
model.eval()
with torch.no_grad():
indices = torch.split(torch.tensor(range(num_docs)), args.eval_batch_size)
eta = get_eta(rnn_inp)
acc_loss = 0
cnt = 0
for idx, ind in enumerate(indices):
data_batch, times_batch = data.get_batch(
tokens, counts, ind, vocab_size, args.emb_size, temporal=True, times=times)
sums = data_batch.sum(1).unsqueeze(1)
if args.bow_norm:
normalized_data_batch = data_batch / sums
else:
normalized_data_batch = data_batch
eta_td = eta[times_batch.type('torch.LongTensor')]
theta = get_theta(eta_td, normalized_data_batch)
print('\n')
print('Get topic coherence...')
print('train_tokens: ', train_tokens[0])
TC_all = []
cnt_all = []
for tt in range(args.num_times):
tc, cnt = get_topic_coherence(beta[:, tt, :].detach().numpy(), train_tokens, vocab)
TC_all.append(tc)
cnt_all.append(cnt)
print('TC_all: ', TC_all)
TC_all = torch.tensor(TC_all)
print('TC_all: ', TC_all.size())
print('\n')
print('Get topic quality...')
quality = tc * diversity
print('Topic Quality is: {}'.format(quality))
print('#'*100)
def get_cluster_quality():
"""Returns cluster quality.
"""
topics_distributions = []
#df = pd.read_csv("/home/beck/Repositories/Data/trec2011_microblog/trec2011_2012_final.tsv", sep="\t")
cluster_ids = []
with open(os.path.join(data_file, 'topic_ids.txt'), 'r') as fp:
for line in fp:
cluster_ids.append(line.strip())
# same number of docs as cluster ids
assert len(cluster_ids) == len(tokens)
rnn_inp = data.get_rnn_input(tokens, counts, times, num_times, vocab_size,
num_docs)
model.eval()
with torch.no_grad():
indices = torch.split(torch.tensor(range(num_docs)),
args.eval_batch_size)
eta = get_eta(rnn_inp)
for idx, ind in enumerate(indices):
data_batch, times_batch = data.get_batch(tokens,
counts,
ind,
vocab_size,
args.emb_size,
temporal=True,
times=times)
sums = data_batch.sum(1).unsqueeze(1)
if args.bow_norm:
normalized_data_batch = data_batch / sums
else:
normalized_data_batch = data_batch
eta_td = eta[times_batch.type('torch.LongTensor')]
theta = get_theta(eta_td, normalized_data_batch)
topics_distributions += torch.argmax(theta, 1).tolist()
for (name, fun) in {
'NMI': NMI,
'Hom': Hom,
'Com': Com,
'VM': VM,
'Acc': Acc
}.items():
print(name, fun(cluster_ids, topics_distributions))
## get data
# 1. vocabulary
print('Getting vocabulary ...')
data_file = os.path.join(args.data_path, 'min_df_{}'.format(args.min_df))
vocab, train, valid, test = data.get_data(data_file, temporal=True)
vocab_size = len(vocab)
args.vocab_size = vocab_size
# 1. training data
print('Getting training data ...')
train_tokens = train['tokens']
train_counts = train['counts']
train_times = train['times']
args.num_times = len(np.unique(train_times))
args.num_docs_train = len(train_tokens)
train_rnn_inp = data.get_rnn_input(
train_tokens, train_counts, train_times, args.num_times, args.vocab_size, args.num_docs_train)
# 2. dev set
print('Getting validation data ...')
valid_tokens = valid['tokens']
valid_counts = valid['counts']
valid_times = valid['times']
args.num_docs_valid = len(valid_tokens)
valid_rnn_inp = data.get_rnn_input(
valid_tokens, valid_counts, valid_times, args.num_times, args.vocab_size, args.num_docs_valid)
# 3. test data
print('Getting testing data ...')
test_tokens = test['tokens']
test_counts = test['counts']
test_times = test['times']
## get data
# 1. vocabulary
print('Getting vocabulary ...')
data_file = os.path.join(args.data_path, 'min_df_{}'.format(args.min_df))
vocab, train_data, validation_data, test_1_data, test_2_data, test_data = data.get_data(
)
vocab_size = len(vocab)
args.vocab_size = vocab_size
# 1. training data
print('Getting training data ...')
_, train_times = data.get_time_columns(train_data)
args.num_times = len(np.unique(train_times))
print("the number of uniques train times are ", args.num_times)
train_rnn_inp = data.get_rnn_input(train_data, args.num_times, args.vocab_size,
"train")
# should save the input here and load it if required
# 2. dev set
print('Getting validation data ...')
valid_rnn_inp = data.get_rnn_input(validation_data, args.num_times,
args.vocab_size, "valid")
# 3. test data
print('Getting testing data ...')
test_rnn_inp = data.get_rnn_input(test_data, args.num_times, args.vocab_size,
"test")
test_1_rnn_inp = data.get_rnn_input(test_1_data, args.num_times,
args.vocab_size, "test_1")
test_2_rnn_inp = data.get_rnn_input(test_2_data, args.num_times,
args.vocab_size, "test_2")