def plot_arun_metric(self, min_num_topics=10, max_num_topics=50, iterations=10):
symmetric_kl_divergence = self.topic_model.arun_metric(min_num_topics, max_num_topics, iterations)
plt.clf()
plt.plot(range(min_num_topics, max_num_topics + 1), symmetric_kl_divergence)
plt.title("Arun et al. metric")
plt.xlabel("number of topics")
plt.ylabel("symmetric KL divergence")
plt.savefig("output/arun.png")
save_topic_number_metrics_data(
"output/arun.tsv", range_=(min_num_topics, max_num_topics), data=symmetric_kl_divergence, metric_type="arun"
)
def plot_greene_metric(self, min_num_topics=10, max_num_topics=20, tao=10, step=5, top_n_words=10):
greene_stability = self.topic_model.greene_metric(
min_num_topics=min_num_topics, max_num_topics=max_num_topics, step=step, top_n_words=top_n_words, tao=tao
)
plt.clf()
plt.plot(np.arange(min_num_topics, max_num_topics + 1, step), greene_stability)
plt.title("Greene et al. metric")
plt.xlabel("number of topics")
plt.ylabel("stability")
plt.savefig("output/greene.png")
save_topic_number_metrics_data(
"output/greene.tsv", range_=(min_num_topics, max_num_topics), data=greene_stability, metric_type="greene"
)
def plot_brunet_metric(self, min_num_topics=10, max_num_topics=50, iterations=10):
cophenetic_correlation = self.topic_model.brunet_metric(min_num_topics, max_num_topics, iterations)
plt.clf()
plt.plot(range(min_num_topics, max_num_topics + 1), cophenetic_correlation)
plt.title("Brunet et al. metric")
plt.xlabel("number of topics")
plt.ylabel("cophenetic correlation coefficient")
plt.savefig("output/brunet.png")
save_topic_number_metrics_data(
"output/brunet.tsv",
range_=(min_num_topics, max_num_topics),
data=cophenetic_correlation,
metric_type="brunet",
)
def plot_brunet_metric(self,
min_num_topics=10,
max_num_topics=50,
iterations=10):
cophenetic_correlation = self.topic_model.brunet_metric(
min_num_topics, max_num_topics, iterations)
plt.clf()
plt.plot(range(min_num_topics, max_num_topics + 1),
cophenetic_correlation)
plt.title('Brunet et al. metric')
plt.xlabel('number of topics')
plt.ylabel('cophenetic correlation coefficient')
plt.savefig('output/brunet.png')
save_topic_number_metrics_data('output/brunet.tsv',
range_=(min_num_topics, max_num_topics),
data=cophenetic_correlation,
metric_type='brunet')
def plot_arun_metric(self,
min_num_topics=10,
max_num_topics=50,
iterations=10):
symmetric_kl_divergence = self.topic_model.arun_metric(
min_num_topics, max_num_topics, iterations)
plt.clf()
plt.plot(range(min_num_topics, max_num_topics + 1),
symmetric_kl_divergence)
plt.title('Arun et al. metric')
plt.xlabel('number of topics')
plt.ylabel('symmetric KL divergence')
plt.savefig('output/arun.png')
save_topic_number_metrics_data('output/arun.tsv',
range_=(min_num_topics, max_num_topics),
data=symmetric_kl_divergence,
metric_type='arun')
def plot_greene_metric(self,
min_num_topics=10,
max_num_topics=20,
tao=10,
step=5,
top_n_words=10):
greene_stability = self.topic_model.greene_metric(
min_num_topics=min_num_topics,
max_num_topics=max_num_topics,
step=step,
top_n_words=top_n_words,
tao=tao)
plt.clf()
plt.plot(np.arange(min_num_topics, max_num_topics + 1, step),
greene_stability)
plt.title('Greene et al. metric')
plt.xlabel('number of topics')
plt.ylabel('stability')
plt.savefig('output/greene.png')
save_topic_number_metrics_data('output/greene.tsv',
range_=(min_num_topics, max_num_topics),
data=greene_stability,
metric_type='greene')