def genMostLikelyMovies(self):
movies = parseMovies()
phi = self.calcPhi()
for topic in xrange(15):
top_movies = np.argsort(phi[:, topic])
print "Topic: %d" % topic
print "\n".join("%s: %.4f" % (movies[movieid][0], phi[movieid, topic]) for movieid in top_movies[-10:])
print ""
def print_most_likely_movies(fname, collection_name):
topic_dist = topic_distribution(fname, collection_name)
movies = parseMovies()
for topic in xrange(topic_dist.shape[1]):
top_movies = np.argsort(topic_dist[:, topic])
print "Topic: %d" % topic
print "\n".join("%s: %.4f" % (movies[movieid][0], topic_dist[movieid, topic]) for movieid in top_movies[-10:])
print ""
def genMostLikelyTopic(self):
phi = self.calcPhi()
movies = parseMovies()
topics = defaultdict(list)
for movieid in xrange(self.info["movies"]):
top_topic = np.argsort(phi[movieid, :])[-1]
topics[top_topic].append((movies[movieid][0],
phi[movieid, top_topic]))
return topics
def genMostLikelyMovies(self):
movies = parseMovies()
phi = self.calcPhi()
for topic in xrange(15):
top_movies = np.argsort(phi[:, topic])
print "Topic: %d" % topic
print "\n".join("%s: %.4f" %
(movies[movieid][0], phi[movieid, topic])
for movieid in top_movies[-10:])
print ""
def genMostLikelyTopic(self):
phi = self.calcPhi()
movies = parseMovies()
topics = defaultdict(list)
for movieid in xrange(self.info["movies"]):
top_topic = np.argsort(phi[movieid, :])[-1]
topics[top_topic].append((movies[movieid][0], phi[movieid,
top_topic]))
return topics
def print_most_likely_movies(fname, collection_name):
topic_dist = topic_distribution(fname, collection_name)
movies = parseMovies()
for topic in xrange(topic_dist.shape[1]):
top_movies = np.argsort(topic_dist[:, topic])
print "Topic: %d" % topic
print "\n".join("%s: %.4f" %
(movies[movieid][0], topic_dist[movieid, topic])
for movieid in top_movies[-10:])
print ""
def visualizePCA(self, samples=20):
phi = self.calcPhi()
movies = parseMovies()
pca = PCA(phi)
indices = sample(xrange(len(movies)), samples)
x_axis = pca.Y[indices, 0]
y_axis = pca.Y[indices, 1]
fig = plt.figure()
fig.set_size_inches(10, 8)
ax = fig.add_subplot(111)
ax.scatter(x_axis, y_axis)
for idx, x, y in izip(indices, x_axis, y_axis):
ax.annotate(movies[idx][0].decode('ascii', 'ignore').encode('ascii', 'ignore'), (x, y))
return fig
def visualizePCA(self, samples=20):
phi = self.calcPhi()
movies = parseMovies()
pca = PCA(phi)
indices = sample(xrange(len(movies)), samples)
x_axis = pca.Y[indices, 0]
y_axis = pca.Y[indices, 1]
fig = plt.figure()
fig.set_size_inches(10, 8)
ax = fig.add_subplot(111)
ax.scatter(x_axis, y_axis)
for idx, x, y in izip(indices, x_axis, y_axis):
ax.annotate(
movies[idx][0].decode('ascii',
'ignore').encode('ascii', 'ignore'),
(x, y))
return fig