Players

Commits

Jul 31, 2012

Merge pull request #983 from GaelVaroquaux/faster_ward MRG: Faster ward

1601b276be341d1dfa05a1ed54c66df91cbbb2a7

Jul 30, 2012

Jul 29, 2012

TST upgrade trivial (single-class) k-NN problems to binary ones XXX should we raise a warning/exception when there is only one class in y?

f01907dcb208edf46dc6b5b652aa46b52c62ba59

Jul 28, 2012

Jul 26, 2012

Merge branch 'master' of github.com:scikit-learn/scikit-learn

6de846622f4e581d442e265eaa5be29e554dbd5d

Merge branch 'linking_arrayfuncs' Conflicts: setup.py sklearn/__init__.py

b5519f08dabcbd632ce930273af6f9265aac5b1b

MISC: build_utils: module rather than a subpackages

446611dbcfda4db0722ccd67020e9059423cc10e

ENH: Faster ward for large n_clusters Stop the tree early. This can be interesting as the last steps in the tree building are actually the most costly.

7a86168d85ef2d4f8a0ead96276d8caae36d4e10

Jul 25, 2012

Merge pull request #1 from cournape/linking_arrayfuncs REF: hack to be able to share distutils utilities.

bb73ff8e17c4d20da6a3433dba9b5041e5e39fdf

REF: hack to be able to share distutils utilities. We took the ugly trick from NumPy to separate build vs normal use of sklearn imports.

d8e30f1ca8dad740e8b936c9c8d3f094cf915810

Jul 24, 2012

ENH corrected errormessages for always present labels. ugh

3ebdf35caf29b4422dbfc47c0d41900e4ce3a89d

BF: adapt a svm test to recent numpy versions. Use assert_raises((Exception1, Exception2), ...) syntax to catch an exception that may vary according to numpy version.

68693b8f4720e6549ca02c268096bbd8a1e888b7

Added test to sklearn.utils.bench's total_seconds method

73a274ee7a63bd38a39afc546d05726cdce59540

Jul 23, 2012

ENH better error messaged in multiclass as @mbondel suggested.

b90a30dd488491efa69dd617c9192d37225eaddd

COSMIT, removed scikits.learn things, removed orphan file.

408d3ef86dc4806aff43cc715c0b034050c621f7

Merge pull request #955 from vene/mem_prof [WIP] Add memory_profiler tips to the developer's guide.

fbcde2d59e2a7894ace705b909792c8bd85c8747

Begin updating the developers performance documentation

09a7d78c77cb7c89f2a6c0ee4bd52e0af65e77cd

Update and fix errors in memory profiling documentation

4c533801b8771943b12927e2890278fdd6d0fe1b

ENH: sklearn.setup_module to preseed RNGs to reproduce failures Could easily be controlled from the environment via SKLEARN_SEED Besides adding setup_module() to the sklearn space should be of no effect for any regular imports of sklearn.

4915d4dcd93a1051f4bdf933987943118d02c9a7

FIX: remove leftover print It was making doctests fail

91503547d168c15635465c84a497faf094dd7e48

Put Ledoit-Wolf shrinkage coefficient estimation in a separate function.

2e5f874f127d8837645a1b97f16fd58f3f909786

TST: test on diabetes rather than iris We want more features than the block_size in ledoit_wolf_shrinkage, elsewhere we are not really testing this codepath

3fdd1e35af9e9dfd47050aa2e51610003e4b22fa

ENH: make LedoitWolf estimation scale (memory usage) with n_features.

53b4e93a503273b89997a572d03d96a12294dc76

ENH: using slices rather than indice masks Avoid fancy indexing, and thus copies

6aa7450f616b1d7e3222acbe8e3ca0082f73f138

Avoid extra computations + clean `assume_centered` argument use.

7b34256672697dbd2e43131471aa0fb636432893

rename mahalanobis_values to raw_values in covariance decision method. This is to ensure further compatibility with other classifiers' decision function. An optional argument was also added to mahalanobis_distance method in order to consider already centered data.

c24702e09a644443139735805048121298d71416

Catch non-invertibility errors within MinCovDet computation.

c9d7c0546c1c9e84e33a19d418804d3c72293a43

The LedoitWolf object has to return a covariance estimate or breaks.

f3fc22963318389d758a52de80e7639d6e1ea83d

Merge pull request #886 from amueller/multiclass_always_present MRG: enh in OvR, use constant predictor if one class always present or never.

d9226d0f9e62f3b56ce28c5eb3c6a7d6f79c81f2

This test repeats issues 960, with the silhouette coefficient returning nan

5a43dfa9e44034ffcccc356849fdb4c115da30df

Distance matrix doesn't matter, and was therefore removed

91c6c3b84c1ab90f28c63a2dd8b09542ac789f1c

Merge pull request #974 from jaquesgrobler/master Fixes broken links on `Support` page of website

b6c8776c2f50c97450e22bc533c96161c1d436c6

ENH allow user-specified comment in SVMlight dumper

35b24cac4ce8e9187d5f0aa66f401bbb55d0355c

Jul 22, 2012

Merge branch 'master' of https://github.com/scikit-learn/scikit-learn

b9098ef894b812863d5b0ee8239bf249c828294a

DOC doctest failed on my box because I had higher precision...

180312b7f90f83b03d2df4eed049e59e10db3ced

Jul 21, 2012

Jul 20, 2012

ENH put atlas checking in one place and load from there.

c1c734af47a4ba9df81da6325f7457a08e7ac655

Merge pull request #971 from acompa/master Issue #339: minor optimization

bbd1350f75e90953ab55f56b90cd40f0701dd23a

Issue #339: minimizing number of calls in tests.test_hmm. Profiling shows tests.test.train_hmm_and_keep_track_of_log_likelihood takes 3s to run. This decreases that number a bit.

a30da06467c013ac79c48166bda7e3d815954801

ENH classes_ on OvR classifier Getting to the classes via label_binarizer_ violates the Law of Demeter: "Only talk to your immediate friends." (https://en.wikipedia.org/wiki/Law_of_Demeter)

a4845351afe906e74c8f316f12655b7fdc36deb4

COSMIT + DOC input handling and docstrings in RandomizedPCA

14260878472eba102673dadef9aa2fbb06d19753

Jul 19, 2012

Merge pull request #969 from jaquesgrobler/master Documentation fix - BaseLibSVM

f397adbee716536948e4a822fe387136712e5880

Merge branch 'master' of https://github.com/scikit-learn/scikit-learn

57d49e75260a7af53130373c5576542e949129dd

Jul 18, 2012

Merge pull request #946 from glouppe/tree-speedup [MRG] Tree speedup

a2bb8f7c9f98bff4042ff1e4d33f96f16d213534