2d734a1ff7
FIX: broken benchmark
2011-12-19 03:53:53 -05:00
a33485f34d
Remove useless benchmark.
...
It seems incomplete, it doesn't measure anything. Furthermore, git
blame points to me, so I assume I'm not making anybody angry.
2011-12-19 03:48:10 -05:00
36d0adb8c1
pep8 compliant
2011-12-17 00:48:40 +05:30
ae3b9439ca
Re-ran the tests and found that the GaussianNB error was much lower.
2011-11-29 11:58:12 +00:00
a0c841419b
Benchmark: Added random forests and extra-trees to bench_sgd_covertype.py
2011-11-23 13:15:19 +01:00
26b5896aaa
Delette benchmarks/bench_neighbors.py
...
ml-benchmarks should be used instead.
2011-10-25 13:59:42 +02:00
634cf71631
Delete benchmarks/bench_svm.py
...
ml-benchmarks has a much more comprehensive benchmark.]
2011-10-25 13:58:52 +02:00
6182960ac9
Merge branch 'master' of github.com:scikit-learn/scikit-learn into enh/tree
2011-09-21 20:26:40 +01:00
097f26a297
we grow our trees deep
2011-09-16 17:34:13 +02:00
c74347f5ca
trailing spaces
2011-09-16 11:29:26 +02:00
cca5aff0a2
cosmit: pep8 + docs
2011-09-12 21:33:23 +02:00
53be9c91ce
benchmark -> bar plot
2011-09-11 22:33:04 -07:00
dfbc00dd1d
pep8 + move weighted_mode to utils
2011-09-11 18:35:49 -07:00
3675291ff9
add graphical neighbors benchmark
2011-09-11 18:16:08 -07:00
a170da3264
Merge remote branch 'bdholt1/enh/tree' into bdholt1-enh/tree
2011-09-09 16:13:31 +02:00
9c9320dbc2
time is measured in seconds
2011-09-08 18:49:48 +01:00
9f42fd3b9c
added CART to covertype benchmark -> look at that error rate!
2011-09-07 20:52:28 +02:00
0718f7231b
merge upstream/master
2011-09-06 20:14:25 -07:00
78dd8b5626
batch trailing spaces cleanup
2011-09-06 23:12:26 +02:00
db24cfdf4b
add neighbors benchmark
2011-09-04 18:16:55 -07:00
f0f0182e34
More style consistency improvements
2011-09-04 10:54:21 +02:00
9b5c7cdf56
more renamings
2011-09-03 20:58:20 +02:00
b484aa4432
merged upstream-master into enh/tree
2011-09-03 13:04:27 +01:00
a57c2ae933
more scikits.learn => sklearn updates
2011-09-03 12:57:53 +02:00
b5d63bafe2
renamed K to n_classes
2011-09-03 10:35:07 +01:00
d0e4757855
Removed unnecessary import
2011-08-22 22:08:51 +01:00
73fc08a12d
Merge git://github.com/scikit-learn/scikit-learn into enh/tree
2011-08-22 21:33:07 +01:00
097c3d23a0
ENH: More interesting benchmarks for OMP
2011-08-20 21:34:33 +02:00
0690a0a545
Merge https://github.com/scikit-learn/scikit-learn into enh/tree
2011-08-11 18:48:40 +01:00
ebe226de75
Updated benchmarking for trees
2011-08-09 12:45:53 +01:00
36c18578ad
Merge remote-tracking branch 'upstream/master' into samples_generator
...
Resolved conflicts in:
scikits/learn/datasets/__init__.py
scikits/learn/datasets/samples_generator.py
2011-08-06 17:01:48 +02:00
2cd0ccf71e
updated samples generator according to @glouppe's refactoring
2011-08-04 16:50:47 +03:00
3fd8292c18
Fixes for broken benchmarks due to API changes in samples_generator.py
2011-08-04 15:09:32 +02:00
5bca101786
ENH : prettify OMP/LARS benchmark
2011-08-03 23:00:48 -04:00
d4eadd7d85
clearer benchmark
2011-08-03 21:21:58 +03:00
c018c5fdd8
FIX: benchmark was broken
2011-08-03 17:09:42 +03:00
b711769e26
pep8
2011-08-03 17:04:26 +03:00
6af2d2a345
removed occurances of tree_model
2011-08-02 16:56:54 +01:00
110ae3cb3d
while 1 instead of while True
2011-08-02 18:56:44 +03:00
7c43e7d7c8
use np.asfortranarray
2011-08-02 13:08:47 +03:00
56b1df8093
warn when omp stops early
2011-08-02 00:40:05 +03:00
3de1c0ab9e
resolved conflicts
2011-07-29 12:00:30 -04:00
6a8ce6f7e8
make number of classes explicit
2011-07-29 13:58:44 +01:00
24348af224
Refactored decision trees and forests to support CART algorithm.
...
Notable changes:
1) Supports classification and regression
2) 3 classification criteria, 1 regression criterion
3) A new dataset is provided to test regression (Boston House Prices)
4) Weights are removed from the algorithm entirely. If the need for weights can be justified, I would welcome reintroducing them, but for the refactoring I left them out.
5) The subset of dimensions (F) to split on is fixed for the entire tree, not at each node. This is more in line with CART and RandomForests.
6) A max_depth parameter is offered to limit the size of the constructed trees.
7) Randomisation is fixed with python's random module, but can be seeded.
8) For classification, the number of classes must be provided when the tree is constructed. This is because the tree cannot necessarily infer the correct number of labels at the time of training if only a subset of the data is used for individual trees.
9) For classification, labels are not normalised internally. Labels must be provided to the algorithm in the range [0, ..., K)
10) For classification, the leaf nodes retain the distribution of classes. This means that it is possible to query the tree for the probability distribution of a test sample
2011-07-29 12:20:03 +01:00
7435b1ddfb
Added OMP bench
2011-07-29 14:06:35 +03:00
6134371d54
MISC: s/larslasso_results/lars_lasso_results
2011-07-26 09:52:32 -04:00
5a12763441
API : renaming LARS to Lars
2011-07-25 21:41:48 -04:00
ccacfb382b
Cosmit: Fix (some) pyflakes warnings
2011-07-16 23:53:10 +02:00
943439d4b0
rename MultinomialNB params, rename GNB GaussianNB
2011-05-23 16:21:44 +02:00
cebecf8335
[MiniBatchKMeans] Added labels and scaled the axis properly on the benchmark plot
2011-05-17 20:46:55 +01:00
Fabian Pedregosa