Merge a084954 into 702a802

mvpa/cosmo_cartprod.m

@@ -84,7 +84,7 @@
     p=cartprod(xs);
 
     % if input was a struct, output is a cell with structs
-    if as_struct();
+    if as_struct
         p=cell2structs(p, fns);
     elseif convert_to_numeric && ~isempty(p) && ...
                         all(cellfun(@isnumeric,p(:)))

mvpa/cosmo_check_external.m

@@ -162,7 +162,7 @@
 
 function is_ok=check_external_toolbox(external_name,raise_)
     externals=get_externals();
-    if ~isfield(externals, external_name);
+    if ~isfield(externals, external_name)
         error('Unknown external ''%s''', external_name);
     end
 
@@ -292,10 +292,6 @@
 
     tf=cached_tf;
 
-    %paths=cosmo_strsplit(path(),pathsep());
-    %starts_with_toolbox_dir=@(x)isempty(cosmo_strsplit(x,toolbox_dir,1));
-    %f=any(cellfun(starts_with_toolbox_dir,paths));
-
 
 function s=url2str(url)
     if strcmp(cosmo_wtf('environment'),'matlab')
@@ -405,6 +401,18 @@
                               'Article ID 156869, 9 pages.',...
                               'doi:10.1155/2011/156869'];
 
+    externals.eeglab.is_present=@() has('eeglab') ...
+                                       && has('pop_loadset');
+    externals.eeglab.is_recent=yes;
+    externals.eeglab.label='EEGLAB toolbox';
+    externals.eeglab.url='https://sccn.ucsd.edu/eeglab/';
+    externals.eeglab.authors={'A. Delorme','S. Makeig'};
+    externals.eeglab.year='2004';
+    externals.eeglab.ref=['EEGLAB: an open source toolbox for analysis '...
+                            'of single-trial EEG dynamics. '...
+                            'Journal of Neuroscience Methods 134:9-21'];
+
+
     externals.libsvm.is_present=@() has('svmpredict') && ...
                                         has('svmtrain');
     % require version 3.18 or later, because it has a 'quiet' option

mvpa/cosmo_oddeven_partitioner.m

@@ -4,8 +4,7 @@
 % partitions=cosmo_oddeven_partitioner(chunks,[type])
 %
 % Input
-%    ds              dataset struct with field .ds.chunks, or Px1 chunk
-%                    indices (for P samples).
+%    ds              dataset struct with field .ds.chunks.
 %    type            One of:
 %                    - 'full': two partitions are returned, training on odd
 %                       unique chunks and testing on even unique chunks,
@@ -30,7 +29,8 @@
 %
 % Example:
 %     ds=struct();
-%     ds.sa.samples=NaN(6,99); % will be ignored by this function
+%     ds.samples=NaN(8,99); % will be ignored by this function
+%     ds.sa.targets=[8 9 8 9 8 9 8 9]';
 %     ds.sa.chunks=[1 1 2 2 6 7 7 6]';
 %     p=cosmo_oddeven_partitioner(ds);
 %     % note that chunks=6 ends up in the odd chunks and chunks=7 in the
@@ -133,7 +133,8 @@
     partitions.train_indices=train_indices;
     partitions.test_indices=test_indices;
 
-
+    % final check to make sure all is kosher (including balanced-ness)
+    cosmo_check_partitions(partitions,ds);
 
 function ds=get_chunks(ds)
     if isnumeric(ds) && isvector(ds)

tests/test_meeg_io.m

@@ -396,7 +396,7 @@ function test_eeglab_io_ersp()
                          {'ersp'},...
                          {false,true}});
 
-     ncombi=size(args,1);
+    ncombi=size(args,1);
     for k=1:ncombi
         arg=args(k,1:3);
         [s,ds_cell,ext]=build_eeglab_dataset_struct(arg{:});
@@ -558,7 +558,7 @@ function test_eeglab_io_exceptions()
         case 'erspbase'
             chan_suffix='_erspbase';
 
-        case 'itc';
+        case 'itc'
             chan_suffix='_itc';
 
         otherwise
@@ -577,7 +577,7 @@ function test_eeglab_io_exceptions()
             has_freq=true;
 
 
-        case {'erp'};
+        case {'erp'}
             has_freq=false;
 
         otherwise
@@ -602,7 +602,7 @@ function test_eeglab_io_exceptions()
     if hastime
         % include time dimension
         time_label={'time'};
-        time_value={(1:time_dim())*.2-.1};
+        time_value={(1:time_dim)*.2-.1};
     else
         % no time dimension
         time_dim=[];

tests/test_oddeven_partitioner.m

@@ -10,11 +10,11 @@
     initTestSuite;
 
 function test_test_oddeven_partitioner_basics
-    for nchunks=[2 6 7 8]
+    for nchunks=[2 6 8]
         ds=cosmo_synthetic_dataset('ntargets',3,'nchunks',nchunks);
 
         nsamples=size(ds.samples,1);
-        rp=randperm(ceil(nsamples*.7));
+        rp=randperm(nsamples);
         ds=cosmo_slice(ds,[rp rp]);
 
         unq_chunks=unique(ds.sa.chunks);
@@ -27,15 +27,15 @@
         fp.train_indices={idx_odd, idx_even};
         fp.test_indices={idx_even, idx_odd};
         assert_partitions_equal(fp,cosmo_oddeven_partitioner(ds,'full'));
-        c=ds.sa.chunks;
-        assert_partitions_equal(fp,cosmo_oddeven_partitioner(c,'full'));
+        %c=ds.sa.chunks;
+        %assert_partitions_equal(fp,cosmo_oddeven_partitioner(c,'full'));
 
         hp=struct();
         hp.train_indices={idx_odd};
         hp.test_indices={idx_even};
         assert_partitions_equal(hp,cosmo_oddeven_partitioner(ds,'half'));
-        c=ds.sa.chunks;
-        assert_partitions_equal(hp,cosmo_oddeven_partitioner(c,'half'));
+        %c=ds.sa.chunks;
+        %assert_partitions_equal(hp,cosmo_oddeven_partitioner(c,'half'));
     end
 
 
@@ -69,7 +69,26 @@ function assert_cell_same_elements(p,q)
         assertEqual(sort(p{k}),sort(q{k}));
     end
 
-
+function test_unbalanced_partitions()
+    ds=struct();
+    for ntargets=2:5
+        for nchunks=2:5
+            nsamples=ntargets*nchunks;
+            ds.samples=randn(nsamples,1);
+            ds.sa.chunks=repmat(1:nchunks,1,ntargets)';
+            ds.sa.targets=ceil((1:nsamples)'/nchunks);
+
+            idxs=cosmo_randperm(numel(ds.sa.chunks));
+            ds=cosmo_slice(ds,idxs);
+
+            % should be ok
+            cosmo_oddeven_partitioner(ds);
+
+            idx=ceil(rand()*ntargets*nchunks);
+            ds.sa.chunks(idx)=ds.sa.chunks(idx)+1;
+            assertExceptionThrown(@() cosmo_oddeven_partitioner(ds),'*');
+        end
+    end