Skip to content

[WIP] ENH Optimal n_clusters values #6948

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 5 commits into from
Closed

[WIP] ENH Optimal n_clusters values #6948

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
2fd375c
add distortion
Mar 13, 2015
733390a
[WIP] ENH Add the OptimalNClusterSearch class [ci skip]
tguillemot Jun 29, 2016
227b30d
Add plot_choose_n_optimal example [ci skip]
tguillemot Jun 29, 2016
a0926db
Modify little stuff for testing.
tguillemot Jan 9, 2017
1868d75
[ci-skip] Add doc, unsupervised metric for any clustersearch.
tguillemot Jan 9, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
81 changes: 81 additions & 0 deletions examples/cluster/plot_optimal_n_clusters.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,81 @@
"""Plot the results of the gap criterium."""

# Authors: Thierry Guillemot <thierry.guillemot.work@gmail.com>

import time
import numpy as np
import matplotlib.pyplot as plt

from sklearn.cluster import KMeans, SpectralClustering, OptimalNClusterSearch
from sklearn.datasets import make_blobs, make_circles, make_moons
from sklearn.metrics import calinski_harabaz_score, fowlkes_mallows_score
from sklearn.metrics import silhouette_score
from sklearn.utils import check_random_state


n_samples, n_features, random_state = 1000, 2, 1
parameters = {'n_clusters': np.arange(1, 8)}

rng = check_random_state(random_state)
datasets = [
('3 clusters', make_blobs(n_samples=n_samples, n_features=2,
random_state=random_state, centers=3)),
('5 clusters', make_blobs(n_samples=n_samples, n_features=2,
random_state=random_state, centers=5)),
('random', (rng.rand(n_samples, n_features),
np.zeros(n_samples, dtype=int))),
('circle', make_circles(n_samples=n_samples, factor=.5, noise=.05,
shuffle=True, random_state=random_state)),
('moon', make_moons(n_samples=n_samples, noise=.05,
shuffle=True, random_state=random_state)),
]

estimator = SpectralClustering(eigen_solver='arpack',
affinity="nearest_neighbors",
n_init=10, random_state=0)
# estimator = KMeans(n_init=10, random_state=0)
searchers = [
# ('Silhouette', OptimalNClusterSearch(
# estimator=estimator, parameters=parameters,
# fitting_process='unsupervised', metric=silhouette_score)),
# ('Calinski', OptimalNClusterSearch(
# estimator=estimator, parameters=parameters,
# fitting_process='unsupervised', metric=calinski_harabaz_score)),
# ('Stability', OptimalNClusterSearch(
# estimator=estimator, parameters=parameters, random_state=0,
# fitting_process='stability', metric=fowlkes_mallows_score)),
# ('Distortion jump', OptimalNClusterSearch(
# estimator=estimator, parameters=parameters,
# fitting_process='distortion_jump')),
# ('Gap', OptimalNClusterSearch(
# estimator=estimator, parameters=parameters, random_state=0,
# fitting_process='gap', metric=silhouette_score)),
('Pham', OptimalNClusterSearch(
estimator=estimator, parameters=parameters, fitting_process='pham',
metric=silhouette_score)),
]

color = 'bgrcmyk'
plt.figure(figsize=(13, 9.5))
plt.subplots_adjust(left=.001, right=.999, bottom=.001, top=.96, wspace=.05,
hspace=.01)
for k, (data_name, data) in enumerate(datasets):
X, _ = data
for l, (search_name, search) in enumerate(searchers):
t0 = time.time()
y = search.fit(X).best_estimator_.labels_
t1 = time.time()

colors = np.array([color[k] for k in y])
plt.subplot(len(datasets), len(searchers),
len(searchers) * k + l + 1)
if k == 0:
plt.title(search_name, size=18)
plt.scatter(X[:, 0], X[:, 1], color=colors, alpha=.6)
plt.xticks(())
plt.yticks(())
plt.axis('equal')
plt.text(.99, .01, ('%.2fs' % (t1 - t0)).lstrip('0'),
transform=plt.gca().transAxes, size=15,
horizontalalignment='right')
plt.show()
4 changes: 3 additions & 1 deletion sklearn/cluster/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
from .dbscan_ import dbscan, DBSCAN
from .bicluster import SpectralBiclustering, SpectralCoclustering
from .birch import Birch
from .optimal_nclusters_search import OptimalNClusterSearch

__all__ = ['AffinityPropagation',
'AgglomerativeClustering',
Expand All @@ -33,4 +34,5 @@
'spectral_clustering',
'ward_tree',
'SpectralBiclustering',
'SpectralCoclustering']
'SpectralCoclustering',
'OptimalNClusterSearch']
Loading