Merge pull request #765 from lilleswing/k_fold

    train_ds_base = None

    train_datasets = []

    # rem_dataset is remaining portion of dataset

    if isinstance(dataset, DiskDataset):

      rem_dataset = dataset

    else:

      rem_dataset = DiskDataset.from_numpy(dataset.X, dataset.y, dataset.w,

                                           dataset.ids)

    for fold in range(k):

      # Note starts as 1/k since fold starts at 0. Ends at 1 since fold goes up

      # to k-1.

          frac_train=frac_fold,

          frac_valid=1 - frac_fold,

          frac_test=0)

      cv_dataset = rem_dataset.select(fold_inds)

      cv_dataset = rem_dataset.select(fold_inds, select_dir=cv_dir)

      cv_datasets.append(cv_dataset)

      rem_dataset = rem_dataset.select(rem_inds)

      update_train_base_merge = filter(lambda x: x is not None,

                                       [train_ds_base, cv_dataset])

      train_ds_base = DiskDataset.merge(

          update_train_base_merge, merge_dir=cv_dir)

      train_ds_base = DiskDataset.merge(update_train_base_merge)

    return list(zip(train_datasets, cv_datasets))

  def train_valid_test_split(self,

    fold_datasets = index_splitter.k_fold_split(ds, K)

    for fold in range(K):

      self.assertTrue(fold_datasets[fold][1].X[0] == fold)

      train_data = set(list(fold_datasets[fold][0].X))

      train, cv = fold_datasets[fold][0], fold_datasets[fold][1]

      self.assertTrue(cv.X[0] == fold)

      train_data = set(list(train.X))

      self.assertFalse(fold in train_data)

      self.assertEqual(K - 1, len(train_data))

      self.assertEqual(K - 1, len(train))

      self.assertEqual(1, len(cv))

  def test_singletask_random_k_fold_split(self):

    """