def main(fpath, vocab_fpath, num_cluster, pca):
song_matrix = read_sparse(fpath)
vocab = read_vocabuary(vocab_fpath)
data = None
if not pca:
data = song_matrix
else:
pca_algo = RandomizedPCA(n_components=pca)
data = pca_algo.fit(song_matrix).transform(song_matrix)
labels_column = run_n_kmeans(data, num_cluster, num_cluster, 2)
_summarize(song_matrix, vocab, labels_column, num_cluster)
def main(fpath):
song_matrix = read_sparse(fpath)
to_plot_x = [0] + range(2, 21) + [25, 35]
to_plot_y = [0]
errors = [0]
for x in to_plot_x[1:]:
print(x)
beta_cv, err = kmeans_betacv(song_matrix, x, True)
to_plot_y.append(beta_cv)
errors.append(err)
ax = plt.gca()
ax.set_xticks(to_plot_x)
plt.ylabel('IntraCluster/InterCluster Ratio')
plt.xlabel('Number of clusters')
plt.errorbar(to_plot_x, to_plot_y, yerr=errors,
fmt='bo', markersize=8, elinewidth=2)
plt.show()
def main(fpath, hex):
song_matrix = read_sparse(fpath)
pca = RandomizedPCA(n_components = 2)
pcas = pca.fit(song_matrix).transform(song_matrix)
print(pca.explained_variance_ratio_)
if hex:
plt.hexbin(pcas[:,0], pcas[:,1], cmap=cm.get_cmap('bone_r', 100),
bins='log', gridsize=100, mincnt=2)
plt.colorbar()
else:
plt.scatter(pcas[:,0], pcas[:,1])
plt.legend()
ax = plt.gca()
plt.ylabel('Second Principal Component')
plt.xlabel('First Principal Component')
plt.title('PCA of the LastFM dataset')
plt.show()