Add kmeans page to cookbook #3183

OXPHOS · 2016-05-05T05:34:20Z

There's no parameter CLabels in class CKMeans, so I can't find a way to use apply_*, eval.evaluate, so as to compare test and training dataset.
Though I didn't see why CKMeans cannot have CLabels - we can just label the clusters 1..N.
I thought about evaluating the clustering performance by calculating the Euclidean distances between centers of training dataset and test dataset, but there's no handy method for now.
I didn't see the difference between dataset fm_train_real.dat and classifier_binary__2d_linear_features_train.datbut I think it doesn't really matter which one to use..?

OXPHOS · 2016-05-05T05:48:46Z

Some general issues -

I ran into the same problem as in OSX Segmentation fault: 11 #3029 when I tried to run clustering_kmeans_modular.py. Import of modshogun resulted in segmentation fault 11 and I had no idea why.
I was trying to translate the meta scripts to other languages. I did ./generate.py --input examples --output outputs, but was told ctags cannot be NoneType. The package ctags can be found in cmake: Ctags flavour: Exuberant (?). I haven't looked into this problem yet.

lisitsyn · 2016-05-05T21:41:22Z

@OXPHOS I will take a look at ctags issue. This was actually caused by me.

Saurabh7 · 2016-05-06T06:03:35Z

@OXPHOS

Have you tried apply_multiclass() , should work for KMeans i think. https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/machine/DistanceMachine.cpp#L208
About calculating the distances to closest centers, I thought it should be possible too. I was actually planning to add it after PR refactor KMeans #3166

OXPHOS · 2016-05-11T06:18:45Z

@lisitsyn Hey the error message is interestingly different from the last time I submitted but I guess it still has something to do with ctags...? I cannot make meta-examples at all now.

Please see below for details. Thanks!

Built target ctags
Generating examples from meta-language
Traceback (most recent call last):
  File "/Users/zora/Github/shogun/examples/meta/generator/generate.py", line 118, in <module>
    includedTargets=args.targets, storeVars=storeVars)
  File "/Users/zora/Github/shogun/examples/meta/generator/generate.py", line 58, in translateExamples
    storeVars=storeVars if target['FileExtension']==".cpp" else False)
  File "/Users/zora/Github/shogun/examples/meta/generator/translate.py", line 365, in translate
    return translator.translateProgram(ast["Program"], programName, tags, storeVars)
  File "/Users/zora/Github/shogun/examples/meta/generator/translate.py", line 56, in translateProgram
    dependencies=self.dependenciesString(),
  File "/Users/zora/Github/shogun/examples/meta/generator/translate.py", line 121, in dependenciesString
    allClassDependencies = template.substitute(classlist=self.seperatedClassDependencies("AllClasses"))
  File "/Users/zora/Github/shogun/examples/meta/generator/translate.py", line 159, in seperatedClassDependencies
    csdependencies += elementTemplate.substitute(element=x, include=self.getIncludePathForClass(x))
  File "/Users/zora/Github/shogun/examples/meta/generator/translate.py", line 181, in getIncludePathForClass
    raise Exception('Failed to obtain include path for %s' % (' or '.join(variants)))
Exception: Failed to obtain include path for int32_t or Cint32_t or int32_t or Cint32_t
make[3]: *** [examples/meta/CMakeFiles/meta_examples] Error 1
make[2]: *** [examples/meta/CMakeFiles/meta_examples.dir/all] Error 2
make[1]: *** [doc/cookbook/CMakeFiles/cookbook.dir/rule] Error 2

karlnapf · 2016-05-13T06:35:02Z

doc/cookbook/source/examples/classifier/kmeans.rst

@@ -0,0 +1,47 @@
+==================
+:math:`k`-means clustering


Minors: Capital K please. Also, can you avoid the math in the header (only there)

OXPHOS · 2016-05-15T05:03:20Z

Error:

 28/32 Test #29: generated_ruby-classifier-kmeans ................................***Failed    0.10 sec
 /opt/shogun/build/examples/meta/ruby/classifier/kmeans.rb:28:in `<main>': undefined method `set_mbKMeans_params' for #<Modshogun::KMeans:0x0000000188fb98> (NoMethodError)

karlnapf · 2016-05-16T14:13:08Z

Interesting. Can you reproduce this locally?

OXPHOS · 2016-05-17T05:01:54Z

I did cmake -DRubyModular=on .. and make.
And in build/examples/meta/ I ran make test, I failed all ruby examples without a given reason, while all cpp examples passed.

Also I tried python modular and they failed too. Did I miss anything?

karlnapf · 2016-05-18T13:33:20Z

What error are they failing on?

karlnapf · 2016-05-18T13:33:41Z

You can also run the meta examples manually. What happens then?

OXPHOS · 2016-05-19T15:40:56Z

I don't think any specific info is provided. Do you mind showing me how to run meta example manually? Thanks.

Run make test:

Running tests...
Test project /Users/zora/Github/shogun/build/examples/meta
      Start  1: meta-examples-generator
 1/19 Test  #1: meta-examples-generator ...........................................   Passed    2.37 sec
      Start  2: generated_cpp-gaussian_naive_bayes
 2/19 Test  #2: generated_cpp-gaussian_naive_bayes ................................   Passed    0.09 sec
      Start  3: generated_cpp-knn
 3/19 Test  #3: generated_cpp-knn .................................................   Passed    0.05 sec
      Start  4: generated_cpp-linear_svm
 4/19 Test  #4: generated_cpp-linear_svm ..........................................   Passed    0.06 sec
      Start  5: generated_cpp-gaussian_process_regression
 5/19 Test  #5: generated_cpp-gaussian_process_regression .........................   Passed    0.21 sec
      Start  6: generated_cpp-kernel_ridge_regression
 6/19 Test  #6: generated_cpp-kernel_ridge_regression .............................   Passed    0.04 sec
      Start  7: generated_cpp-linear_ridge_regression
 7/19 Test  #7: generated_cpp-linear_ridge_regression .............................   Passed    0.04 sec
      Start  8: generated_python-classifier-gaussian_naive_bayes
 8/19 Test  #8: generated_python-classifier-gaussian_naive_bayes ..................***Failed    0.26 sec
      Start  9: generated_python-classifier-knn
 9/19 Test  #9: generated_python-classifier-knn ...................................***Failed    0.04 sec
      Start 10: generated_python-classifier-linear_svm
10/19 Test #10: generated_python-classifier-linear_svm ............................***Failed    0.03 sec
      Start 11: generated_python-gaussian_processes-gaussian_process_regression
11/19 Test #11: generated_python-gaussian_processes-gaussian_process_regression ...***Failed    0.03 sec
      Start 12: generated_python-regression-kernel_ridge_regression
12/19 Test #12: generated_python-regression-kernel_ridge_regression ...............***Failed    0.03 sec
      Start 13: generated_python-regression-linear_ridge_regression
13/19 Test #13: generated_python-regression-linear_ridge_regression ...............***Failed    0.03 sec
      Start 14: generated_ruby-classifier-gaussian_naive_bayes
14/19 Test #14: generated_ruby-classifier-gaussian_naive_bayes ....................***Failed    0.18 sec
      Start 15: generated_ruby-classifier-knn
15/19 Test #15: generated_ruby-classifier-knn .....................................***Failed    0.04 sec
      Start 16: generated_ruby-classifier-linear_svm
16/19 Test #16: generated_ruby-classifier-linear_svm ..............................***Failed    0.04 sec
      Start 17: generated_ruby-gaussian_processes-gaussian_process_regression
17/19 Test #17: generated_ruby-gaussian_processes-gaussian_process_regression .....***Failed    0.05 sec
      Start 18: generated_ruby-regression-kernel_ridge_regression
18/19 Test #18: generated_ruby-regression-kernel_ridge_regression .................***Failed    0.05 sec
      Start 19: generated_ruby-regression-linear_ridge_regression
19/19 Test #19: generated_ruby-regression-linear_ridge_regression .................***Failed    0.04 sec

37% tests passed, 12 tests failed out of 19

Total Test time (real) =   3.77 sec

The following tests FAILED:
      8 - generated_python-classifier-gaussian_naive_bayes (Failed)
      9 - generated_python-classifier-knn (Failed)
     10 - generated_python-classifier-linear_svm (Failed)
     11 - generated_python-gaussian_processes-gaussian_process_regression (Failed)
     12 - generated_python-regression-kernel_ridge_regression (Failed)
     13 - generated_python-regression-linear_ridge_regression (Failed)
     14 - generated_ruby-classifier-gaussian_naive_bayes (Failed)
     15 - generated_ruby-classifier-knn (Failed)
     16 - generated_ruby-classifier-linear_svm (Failed)
     17 - generated_ruby-gaussian_processes-gaussian_process_regression (Failed)
     18 - generated_ruby-regression-kernel_ridge_regression (Failed)
     19 - generated_ruby-regression-linear_ridge_regression (Failed)
Errors while running CTest
make: *** [test] Error 8

karlnapf · 2016-05-21T15:20:14Z

Just run the generated code listing or compiled binary from its directory

On Thursday, 19 May 2016, OXPHOS notifications@github.com wrote:

I don't think any specific info is provided. Do you mind showing me how to
run meta example manually? Thanks.

Run make test:

Running tests...
Test project /Users/zora/Github/shogun/build/examples/meta
Start 1: meta-examples-generator
1/19 Test #1: meta-examples-generator ........................................... Passed 2.37 sec
Start 2: generated_cpp-gaussian_naive_bayes
2/19 Test #2: generated_cpp-gaussian_naive_bayes ................................ Passed 0.09 sec
Start 3: generated_cpp-knn
3/19 Test #3: generated_cpp-knn ................................................. Passed 0.05 sec
Start 4: generated_cpp-linear_svm
4/19 Test #4: generated_cpp-linear_svm .......................................... Passed 0.06 sec
Start 5: generated_cpp-gaussian_process_regression
5/19 Test #5: generated_cpp-gaussian_process_regression ......................... Passed 0.21 sec
Start 6: generated_cpp-kernel_ridge_regression
6/19 Test #6: generated_cpp-kernel_ridge_regression ............................. Passed 0.04 sec
Start 7: generated_cpp-linear_ridge_regression
7/19 Test #7: generated_cpp-linear_ridge_regression ............................. Passed 0.04 sec
Start 8: generated_python-classifier-gaussian_naive_bayes
8/19 Test #8: generated_python-classifier-gaussian_naive_bayes .................._Failed 0.26 sec
Start 9: generated_python-classifier-knn
9/19 Test #9: generated_python-classifier-knn ..................................._Failed 0.04 sec
Start 10: generated_python-classifier-linear_svm
10/19 Test #10: generated_python-classifier-linear_svm ............................_Failed 0.03 sec
Start 11: generated_python-gaussian_processes-gaussian_process_regression
11/19 Test #11: generated_python-gaussian_processes-gaussian_process_regression ..._Failed 0.03 sec
Start 12: generated_python-regression-kernel_ridge_regression
12/19 Test #12: generated_python-regression-kernel_ridge_regression ..............._Failed 0.03 sec
Start 13: generated_python-regression-linear_ridge_regression
13/19 Test #13: generated_python-regression-linear_ridge_regression ..............._Failed 0.03 sec
Start 14: generated_ruby-classifier-gaussian_naive_bayes
14/19 Test #14: generated_ruby-classifier-gaussian_naive_bayes ...................._Failed 0.18 sec
Start 15: generated_ruby-classifier-knn
15/19 Test #15: generated_ruby-classifier-knn ....................................._Failed 0.04 sec
Start 16: generated_ruby-classifier-linear_svm
16/19 Test #16: generated_ruby-classifier-linear_svm .............................._Failed 0.04 sec
Start 17: generated_ruby-gaussian_processes-gaussian_process_regression
17/19 Test #17: generated_ruby-gaussian_processes-gaussian_process_regression ....._Failed 0.05 sec
Start 18: generated_ruby-regression-kernel_ridge_regression
18/19 Test #18: generated_ruby-regression-kernel_ridge_regression ................._Failed 0.05 sec
Start 19: generated_ruby-regression-linear_ridge_regression
19/19 Test #19: generated_ruby-regression-linear_ridge_regression ................._Failed 0.04 sec

37% tests passed, 12 tests failed out of 19

Total Test time (real) = 3.77 sec

The following tests FAILED:
8 - generated_python-classifier-gaussian_naive_bayes (Failed)
9 - generated_python-classifier-knn (Failed)
10 - generated_python-classifier-linear_svm (Failed)
11 - generated_python-gaussian_processes-gaussian_process_regression (Failed)
12 - generated_python-regression-kernel_ridge_regression (Failed)
13 - generated_python-regression-linear_ridge_regression (Failed)
14 - generated_ruby-classifier-gaussian_naive_bayes (Failed)
15 - generated_ruby-classifier-knn (Failed)
16 - generated_ruby-classifier-linear_svm (Failed)
17 - generated_ruby-gaussian_processes-gaussian_process_regression (Failed)
18 - generated_ruby-regression-kernel_ridge_regression (Failed)
19 - generated_ruby-regression-linear_ridge_regression (Failed)
Errors while running CTest
make: *** [test] Error 8

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3183 (comment)

Sent from my phone

OXPHOS · 2016-05-21T17:19:36Z

Aha @karlnapf If this is what you refer to - $ruby build/examples/meta/python/classifier/knn.rb
I have the same modshogun issue

l_require.rb:55:in `require': cannot load such file -- modshogun (LoadError)
    from /System/Library/Frameworks/Ruby.framework/Versions/2.0/usr/lib/ruby/2.0.0/rubygems/core_ext/kernel_require.rb:55:in `require'
    from build/examples/meta/ruby/classifier/knn.rb:1:in `<main>'

karlnapf · 2016-05-23T08:14:27Z

That is weird. Is it the same error as you get on travis?
Also, your Python examples also seem to fail, why is that? Are you sure your build is working fine?

OXPHOS · 2016-05-24T05:38:50Z

@karlnapf So I tried both make install and add path.
Python will crash every time I try to run any example (real crash kind of crash)
Ruby returns the same error as on travis for kmeans.rb

Mac148934:classifier zora$ ruby kmeans.rb 
/Library/Ruby/Site/2.0.0/universal-darwin14/modshogun.bundle: warning: rb_define_const: invalid name `__MAC_10_8' for constant
kmeans.rb:28:in `<main>': undefined method `set_mbKMeans_params' for #<Modshogun::KMeans:0x007fbc051695e0> (NoMethodError)

karlnapf · 2016-05-24T09:20:21Z

Can we discuss this in another thread?

OXPHOS · 2016-05-27T04:04:48Z

@karlnapf This should be moved to clustering as well. What do you think of the ruby issue? You want to open a new thread for it?

karlnapf · 2016-05-27T13:23:37Z

Yes move it!
Is the current verison in the PR producing the problems you mentioend? Then I will close locally and check

karlnapf · 2016-05-27T13:59:33Z

Ill just use this version to check now

OXPHOS · 2016-05-27T14:32:23Z

@karlnapf yes this is the version gives problems. thx.

karlnapf · 2016-06-09T09:41:57Z

doc/cookbook/source/examples/clustering/kmeans.rst

+
+----------
+References
+----------


can you add david barbers book as reference as well?

karlnapf · 2016-06-09T09:45:14Z

Ok travis failure is from gmm. As soon as the minor adjustments in the rst are made, we can (finally!) merge :)

OXPHOS · 2016-06-09T21:26:59Z

@karlnapf can check again :)

OXPHOS · 2016-06-10T02:59:44Z

The R is also failing..

karlnapf · 2016-06-10T16:18:37Z

doc/cookbook/source/examples/clustering/kmeans.rst

+In other words, its objective is to minimize:
+
+.. math::
+   x\argmin_\mathbf{S} \sum_{i=1}^{k}\sum_{\mathbf{x}\in S_k}\left \|\boldsymbol{x} - \boldsymbol{\mu}_i  \right \|^{2}


this x in the beginning should not be there :)

karlnapf · 2016-06-10T16:23:22Z

Some minor adjustments are needed. A rebase against develop should make travis happy (I disabled R meta examples)

OXPHOS · 2016-06-11T06:06:45Z

@karlnapf I tried ~30 times on local and the test can be passed, but it keeps failing on travis..So I'll put fixed_centers() back for now but I don't think we should show that in the cookbook.

karlnapf · 2016-06-12T17:26:19Z

fixed centers obviously changes the algorithm (and the integration test data)
The point I was making: do we need that? If so, you should mention what it does.
Though I preferred if you changed the integration test data and removed it

OXPHOS · 2016-06-14T05:29:20Z

@karlnapf I updated the integration test shogun-toolbox/shogun-data#101
I'll update the meta example once the data is merged. thx!

karlnapf · 2016-06-14T14:53:27Z

Merged :)

karlnapf · 2016-06-14T16:23:44Z

Next time you squash: please put a meaningful commit message.
"Add kmeans cookbook and meta example"

karlnapf · 2016-06-14T16:23:52Z

Waiting for travis and then merging

OXPHOS · 2016-06-14T19:56:09Z

[ 98%] Building CXX object examples/meta/cpp/CMakeFiles/kmeans.dir/clustering/kmeans.cpp.o
/opt/shogun/build/examples/meta/cpp/clustering/kmeans.cpp:31:42: error: use of undeclared identifier 'KMM_LLOYD'
auto kmeans = some<CKMeans>(2, distance, KMM_LLOYD);
                                         ^
/opt/shogun/build/examples/meta/cpp/clustering/kmeans.cpp:44:45: error: use of undeclared identifier 'KMM_MINI_BATCH'
auto kmeans_mb = some<CKMeans>(2, distance, KMM_MINI_BATCH);
                                            ^
2 errors generated.
make[2]: *** [examples/meta/cpp/CMakeFiles/kmeans.dir/clustering/kmeans.cpp.o] Error 1
make[1]: *** [examples/meta/cpp/CMakeFiles/kmeans.dir/all] Error 2
make: *** [all] Error 2

EDIT:
Just recalled KMeans was updated

karlnapf · 2016-06-15T15:29:48Z

Aaaand merging

Add kmeans page to cookbook

OXPHOS force-pushed the cookbook branch 2 times, most recently from dcc6aed to 2f7f73f Compare May 11, 2016 23:47

karlnapf reviewed May 13, 2016
View reviewed changes

OXPHOS force-pushed the cookbook branch from d440745 to deb7d2b Compare May 14, 2016 20:52

karlnapf mentioned this pull request May 28, 2016

renamed kmeans methods to be ruby friendly #3217

Merged

OXPHOS force-pushed the cookbook branch from deb7d2b to db0a773 Compare May 28, 2016 21:20

karlnapf reviewed Jun 9, 2016
View reviewed changes

OXPHOS force-pushed the cookbook branch from b514ebd to ca555db Compare June 9, 2016 19:27

karlnapf reviewed Jun 10, 2016
View reviewed changes

OXPHOS force-pushed the cookbook branch 2 times, most recently from 435a0cb to 5956565 Compare June 11, 2016 03:56

OXPHOS force-pushed the cookbook branch 2 times, most recently from 98e01af to c3c2daa Compare June 14, 2016 15:32

OXPHOS force-pushed the cookbook branch from c3c2daa to 3db7fc8 Compare June 14, 2016 20:49

means cookbook - meta examples and integration test

9950533

OXPHOS force-pushed the cookbook branch from 3db7fc8 to 9950533 Compare June 14, 2016 21:27

karlnapf merged commit 59c003f into shogun-toolbox:develop Jun 15, 2016

OXPHOS deleted the cookbook branch June 22, 2016 22:18

karasikov pushed a commit to karasikov/shogun that referenced this pull request Apr 15, 2017

Merge pull request shogun-toolbox#3183 from OXPHOS/cookbook

2773bae

Add kmeans page to cookbook

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add kmeans page to cookbook #3183

Add kmeans page to cookbook #3183

OXPHOS commented May 5, 2016 •

edited

Loading

OXPHOS commented May 5, 2016 •

edited

Loading

lisitsyn commented May 5, 2016

Saurabh7 commented May 6, 2016

OXPHOS commented May 11, 2016

karlnapf May 13, 2016

OXPHOS commented May 15, 2016 •

edited

Loading

karlnapf commented May 16, 2016

OXPHOS commented May 17, 2016

karlnapf commented May 18, 2016

karlnapf commented May 18, 2016

OXPHOS commented May 19, 2016

karlnapf commented May 21, 2016

OXPHOS commented May 21, 2016

karlnapf commented May 23, 2016

OXPHOS commented May 24, 2016

karlnapf commented May 24, 2016

OXPHOS commented May 27, 2016

karlnapf commented May 27, 2016

karlnapf commented May 27, 2016

OXPHOS commented May 27, 2016

karlnapf Jun 9, 2016

karlnapf commented Jun 9, 2016

OXPHOS commented Jun 9, 2016

OXPHOS commented Jun 10, 2016

karlnapf Jun 10, 2016

karlnapf commented Jun 10, 2016 •

edited

Loading

OXPHOS commented Jun 11, 2016

karlnapf commented Jun 12, 2016

OXPHOS commented Jun 14, 2016

karlnapf commented Jun 14, 2016

karlnapf commented Jun 14, 2016

karlnapf commented Jun 14, 2016

OXPHOS commented Jun 14, 2016 •

edited

Loading

karlnapf commented Jun 15, 2016

		@@ -0,0 +1,47 @@
		==================
		:math:`k`-means clustering

Add kmeans page to cookbook #3183

Add kmeans page to cookbook #3183

Conversation

OXPHOS commented May 5, 2016 • edited Loading

OXPHOS commented May 5, 2016 • edited Loading

lisitsyn commented May 5, 2016

Saurabh7 commented May 6, 2016

OXPHOS commented May 11, 2016

karlnapf May 13, 2016

Choose a reason for hiding this comment

OXPHOS commented May 15, 2016 • edited Loading

karlnapf commented May 16, 2016

OXPHOS commented May 17, 2016

karlnapf commented May 18, 2016

karlnapf commented May 18, 2016

OXPHOS commented May 19, 2016

karlnapf commented May 21, 2016

OXPHOS commented May 21, 2016

karlnapf commented May 23, 2016

OXPHOS commented May 24, 2016

karlnapf commented May 24, 2016

OXPHOS commented May 27, 2016

karlnapf commented May 27, 2016

karlnapf commented May 27, 2016

OXPHOS commented May 27, 2016

karlnapf Jun 9, 2016

Choose a reason for hiding this comment

karlnapf commented Jun 9, 2016

OXPHOS commented Jun 9, 2016

OXPHOS commented Jun 10, 2016

karlnapf Jun 10, 2016

Choose a reason for hiding this comment

karlnapf commented Jun 10, 2016 • edited Loading

OXPHOS commented Jun 11, 2016

karlnapf commented Jun 12, 2016

OXPHOS commented Jun 14, 2016

karlnapf commented Jun 14, 2016

karlnapf commented Jun 14, 2016

karlnapf commented Jun 14, 2016

OXPHOS commented Jun 14, 2016 • edited Loading

karlnapf commented Jun 15, 2016

OXPHOS commented May 5, 2016 •

edited

Loading

OXPHOS commented May 5, 2016 •

edited

Loading

OXPHOS commented May 15, 2016 •

edited

Loading

karlnapf commented Jun 10, 2016 •

edited

Loading

OXPHOS commented Jun 14, 2016 •

edited

Loading