From 9bd0aadaa30295452ba1ba1811b4086cbb2ef5f3 Mon Sep 17 00:00:00 2001
From: Andreas Mueller <amueller@ais.uni-bonn.de>
Date: Sun, 7 Oct 2012 15:13:29 +0200
Subject: [PATCH] MISC renamed n_iterations to n_iter in all other places.
---
benchmarks/bench_plot_svd.py | 18 ++++++-------
.../wikipedia_principal_eigenvector.py | 2 +-
examples/svm/plot_svm_scale_c.py | 2 +-
sklearn/cluster/k_means_.py | 14 +++++------
sklearn/decomposition/pca.py | 2 +-
sklearn/utils/extmath.py | 25 +++++++++++++------
sklearn/utils/tests/test_svd.py | 14 +++++------
7 files changed, 44 insertions(+), 33 deletions(-)
diff --git a/benchmarks/bench_plot_svd.py b/benchmarks/bench_plot_svd.py
index 9f76a23dcc..b8a551c63d 100644
--- a/benchmarks/bench_plot_svd.py
+++ b/benchmarks/bench_plot_svd.py
@@ -12,7 +12,7 @@ from sklearn.utils.extmath import randomized_svd
from sklearn.datasets.samples_generator import make_low_rank_matrix
-def compute_bench(samples_range, features_range, n_iterations=3, rank=50):
+def compute_bench(samples_range, features_range, n_iter=3, rank=50):
it = 0
@@ -36,19 +36,19 @@ def compute_bench(samples_range, features_range, n_iterations=3, rank=50):
results['scipy svd'].append(time() - tstart)
gc.collect()
- print "benching scikit-learn randomized_svd: n_iterations=0"
+ print "benching scikit-learn randomized_svd: n_iter=0"
tstart = time()
- randomized_svd(X, rank, n_iterations=0)
- results['scikit-learn randomized_svd (n_iterations=0)'].append(
+ randomized_svd(X, rank, n_iter=0)
+ results['scikit-learn randomized_svd (n_iter=0)'].append(
time() - tstart)
gc.collect()
- print ("benching scikit-learn randomized_svd: n_iterations=%d "
- % n_iterations)
+ print ("benching scikit-learn randomized_svd: n_iter=%d "
+ % n_iter)
tstart = time()
- randomized_svd(X, rank, n_iterations=n_iterations)
- results['scikit-learn randomized_svd (n_iterations=%d)'
- % n_iterations].append(time() - tstart)
+ randomized_svd(X, rank, n_iter=n_iter)
+ results['scikit-learn randomized_svd (n_iter=%d)'
+ % n_iter].append(time() - tstart)
return results
diff --git a/examples/applications/wikipedia_principal_eigenvector.py b/examples/applications/wikipedia_principal_eigenvector.py
index 0d7bebe691..f053916641 100644
--- a/examples/applications/wikipedia_principal_eigenvector.py
+++ b/examples/applications/wikipedia_principal_eigenvector.py
@@ -172,7 +172,7 @@ names = dict((i, name) for name, i in index_map.iteritems())
print "Computing the principal singular vectors using randomized_svd"
t0 = time()
-U, s, V = randomized_svd(X, 5, n_iterations=3)
+U, s, V = randomized_svd(X, 5, n_iter=3)
print "done in %0.3fs" % (time() - t0)
# print the names of the wikipedia related strongest compenents of the the
diff --git a/examples/svm/plot_svm_scale_c.py b/examples/svm/plot_svm_scale_c.py
index 6b926c4cea..6569e4875c 100644
--- a/examples/svm/plot_svm_scale_c.py
+++ b/examples/svm/plot_svm_scale_c.py
@@ -128,7 +128,7 @@ for fignum, (clf, cs, X, y) in enumerate(clf_sets):
# reduce the variance
grid = GridSearchCV(clf, refit=False, param_grid=param_grid,
cv=ShuffleSplit(n=n_samples, train_size=train_size,
- n_iterations=250, random_state=1))
+ n_iter=250, random_state=1))
grid.fit(X, y)
scores = [x[1] for x in grid.grid_scores_]
diff --git a/sklearn/cluster/k_means_.py b/sklearn/cluster/k_means_.py
index 770eab588f..0f143a5fcf 100644
--- a/sklearn/cluster/k_means_.py
+++ b/sklearn/cluster/k_means_.py
@@ -897,7 +897,7 @@ def _mini_batch_step(X, x_squared_norms, centers, counts,
return inertia, squared_diff
-def _mini_batch_convergence(model, iteration_idx, n_iterations, tol,
+def _mini_batch_convergence(model, iteration_idx, n_iter, tol,
n_samples, centers_squared_diff, batch_inertia,
context, verbose=0):
"""Helper function to encapsulte the early stopping logic"""
@@ -926,7 +926,7 @@ def _mini_batch_convergence(model, iteration_idx, n_iterations, tol,
progress_msg = (
'Minibatch iteration %d/%d:'
'mean batch inertia: %f, ewa inertia: %f ' % (
- iteration_idx + 1, n_iterations, batch_inertia,
+ iteration_idx + 1, n_iter, batch_inertia,
ewa_inertia))
print progress_msg
@@ -935,7 +935,7 @@ def _mini_batch_convergence(model, iteration_idx, n_iterations, tol,
if tol > 0.0 and ewa_diff < tol:
if verbose:
print 'Converged (small centers change) at iteration %d/%d' % (
- iteration_idx + 1, n_iterations)
+ iteration_idx + 1, n_iter)
return True
# Early stopping heuristic due to lack of improvement on smoothed inertia
@@ -952,7 +952,7 @@ def _mini_batch_convergence(model, iteration_idx, n_iterations, tol,
if verbose:
print ('Converged (lack of improvement in inertia)'
' at iteration %d/%d' % (
- iteration_idx + 1, n_iterations))
+ iteration_idx + 1, n_iter))
return True
# update the convergence context to maintain state across sucessive calls:
@@ -1102,7 +1102,7 @@ class MiniBatchKMeans(KMeans):
distances = np.zeros(self.batch_size, dtype=np.float64)
n_batches = int(np.ceil(float(n_samples) / self.batch_size))
- n_iterations = int(self.max_iter * n_batches)
+ n_iter = int(self.max_iter * n_batches)
init_size = self.init_size
if init_size is None:
@@ -1158,7 +1158,7 @@ class MiniBatchKMeans(KMeans):
# Perform the iterative optimization untill the final convergence
# criterion
- for iteration_idx in xrange(n_iterations):
+ for iteration_idx in xrange(n_iter):
# Sample the minibatch from the full dataset
minibatch_indices = self.random_state.random_integers(
@@ -1172,7 +1172,7 @@ class MiniBatchKMeans(KMeans):
# Monitor the convergence and do early stopping if necessary
if _mini_batch_convergence(
- self, iteration_idx, n_iterations, tol, n_samples,
+ self, iteration_idx, n_iter, tol, n_samples,
centers_squared_diff, batch_inertia, convergence_context,
verbose=self.verbose):
break
diff --git a/sklearn/decomposition/pca.py b/sklearn/decomposition/pca.py
index 57c5591396..fdab4a51ca 100644
--- a/sklearn/decomposition/pca.py
+++ b/sklearn/decomposition/pca.py
@@ -485,7 +485,7 @@ class RandomizedPCA(BaseEstimator, TransformerMixin):
n_components = self.n_components
U, S, V = randomized_svd(X, n_components,
- n_iterations=self.iterated_power,
+ n_iter=self.iterated_power,
random_state=self.random_state)
self.explained_variance_ = exp_var = (S ** 2) / n_samples
diff --git a/sklearn/utils/extmath.py b/sklearn/utils/extmath.py
index 6ce09e502f..6293fcbed5 100644
--- a/sklearn/utils/extmath.py
+++ b/sklearn/utils/extmath.py
@@ -4,6 +4,7 @@ Extended math utilities.
# Authors: G. Varoquaux, A. Gramfort, A. Passos, O. Grisel
# License: BSD
+import warnings
import numpy as np
from scipy import linalg
@@ -78,7 +79,8 @@ def safe_sparse_dot(a, b, dense_output=False):
return np.dot(a, b)
-def randomized_range_finder(A, size, n_iterations, random_state=None):
+def randomized_range_finder(A, size, n_iter, random_state=None,
+ n_iterations=None):
"""Computes an orthonormal matrix whose range approximates the range of A.
Parameters
@@ -87,7 +89,7 @@ def randomized_range_finder(A, size, n_iterations, random_state=None):
The input data matrix
size: integer
Size of the return array
- n_iterations: integer
+ n_iter: integer
Number of power iterations used to stabilize the result
random_state: RandomState or an int seed (0 by default)
A random number generator instance
@@ -106,6 +108,10 @@ def randomized_range_finder(A, size, n_iterations, random_state=None):
approximate matrix decompositions
Halko, et al., 2009 (arXiv:909) http://arxiv.org/pdf/0909.4061
"""
+ if n_iterations is not None:
+ warnings.warn("n_iterations was renamed to n_iter for consistency "
+ "and will be removed in 0.16.", DeprecationWarning)
+ n_iter = n_iterations
random_state = check_random_state(random_state)
# generating random gaussian vectors r with shape: (A.shape[1], size)
@@ -117,7 +123,7 @@ def randomized_range_finder(A, size, n_iterations, random_state=None):
# perform power iterations with Y to further 'imprint' the top
# singular vectors of A in Y
- for i in xrange(n_iterations):
+ for i in xrange(n_iter):
Y = safe_sparse_dot(A, safe_sparse_dot(A.T, Y))
# extracting an orthonormal basis of the A range samples
@@ -125,8 +131,8 @@ def randomized_range_finder(A, size, n_iterations, random_state=None):
return Q
-def randomized_svd(M, n_components, n_oversamples=10, n_iterations=0,
- transpose='auto', random_state=0):
+def randomized_svd(M, n_components, n_oversamples=10, n_iter=0,
+ transpose='auto', random_state=0, n_iterations=None):
"""Computes a truncated randomized SVD
Parameters
@@ -142,7 +148,7 @@ def randomized_svd(M, n_components, n_oversamples=10, n_iterations=0,
to ensure proper conditioning. The total number of random vectors
used to find the range of M is n_components + n_oversamples.
- n_iterations: int (default is 0)
+ n_iter: int (default is 0)
Number of power iterations (can be used to deal with very noisy
problems).
@@ -172,6 +178,11 @@ def randomized_svd(M, n_components, n_oversamples=10, n_iterations=0,
* A randomized algorithm for the decomposition of matrices
Per-Gunnar Martinsson, Vladimir Rokhlin and Mark Tygert
"""
+ if n_iterations is not None:
+ warnings.warn("n_iterations was renamed to n_iter for consistency "
+ "and will be removed in 0.16.", DeprecationWarning)
+ n_iter = n_iterations
+
random_state = check_random_state(random_state)
n_random = n_components + n_oversamples
n_samples, n_features = M.shape
@@ -182,7 +193,7 @@ def randomized_svd(M, n_components, n_oversamples=10, n_iterations=0,
# this implementation is a bit faster with smaller shape[1]
M = M.T
- Q = randomized_range_finder(M, n_random, n_iterations, random_state)
+ Q = randomized_range_finder(M, n_random, n_iter, random_state)
# project M to the (k + p) dimensional space using the basis vectors
B = safe_sparse_dot(Q.T, M)
diff --git a/sklearn/utils/tests/test_svd.py b/sklearn/utils/tests/test_svd.py
index 6e5cf6d963..57e437d49e 100644
--- a/sklearn/utils/tests/test_svd.py
+++ b/sklearn/utils/tests/test_svd.py
@@ -69,14 +69,14 @@ def test_randomized_svd_low_rank_with_noise():
# compute the singular values of X using the fast approximate method
# without the iterated power method
- _, sa, _ = randomized_svd(X, k, n_iterations=0)
+ _, sa, _ = randomized_svd(X, k, n_iter=0)
# the approximation does not tolerate the noise:
assert_greater(np.abs(s[:k] - sa).max(), 0.05)
# compute the singular values of X using the fast approximate method with
# iterated power method
- _, sap, _ = randomized_svd(X, k, n_iterations=5)
+ _, sap, _ = randomized_svd(X, k, n_iter=5)
# the iterated power method is helping getting rid of the noise:
assert_almost_equal(s[:k], sap, decimal=3)
@@ -100,14 +100,14 @@ def test_randomized_svd_infinite_rank():
# compute the singular values of X using the fast approximate method
# without the iterated power method
- _, sa, _ = randomized_svd(X, k, n_iterations=0)
+ _, sa, _ = randomized_svd(X, k, n_iter=0)
# the approximation does not tolerate the noise:
assert_greater(np.abs(s[:k] - sa).max(), 0.1)
# compute the singular values of X using the fast approximate method with
# iterated power method
- _, sap, _ = randomized_svd(X, k, n_iterations=5)
+ _, sap, _ = randomized_svd(X, k, n_iter=5)
# the iterated power method is still managing to get most of the structure
# at the requested rank
@@ -125,11 +125,11 @@ def test_randomized_svd_transpose_consistency():
effective_rank=rank, tail_strength=0.5, random_state=0)
assert_equal(X.shape, (n_samples, n_features))
- U1, s1, V1 = randomized_svd(X, k, n_iterations=3, transpose=False,
+ U1, s1, V1 = randomized_svd(X, k, n_iter=3, transpose=False,
random_state=0)
- U2, s2, V2 = randomized_svd(X, k, n_iterations=3, transpose=True,
+ U2, s2, V2 = randomized_svd(X, k, n_iter=3, transpose=True,
random_state=0)
- U3, s3, V3 = randomized_svd(X, k, n_iterations=3, transpose='auto',
+ U3, s3, V3 = randomized_svd(X, k, n_iter=3, transpose='auto',
random_state=0)
U4, s4, V4 = linalg.svd(X, full_matrices=False)
--
GitLab