04f4305931bedd43ca76190b25de120239d2dfe6,deepchem/splits/tests/test_splitter.py,TestSplitter,test_multitask_stratified_column_indices_masked,#TestSplitter#,381

Before Change

    w[:n_samples // 2] = 0

    stratified_splitter = dc.splits.RandomStratifiedSplitter()
    split_indices = stratified_splitter.get_task_split_indices(
        y, w, frac_split=.5)

    w_present = (w != 0)
    y_present = y * w_present
    for task in range(n_tasks):
      split_index = split_indices[task]
      task_actives = np.count_nonzero(y_present[:, task])
      // The split index should partition dataset in half.
      assert np.count_nonzero(y_present[:split_index, task]) == int(
          task_actives / 2)

After Change

    dataset = dc.data.NumpyDataset(X, y, w)

    stratified_splitter = dc.splits.RandomStratifiedSplitter()
    train, valid, test = stratified_splitter.split(dataset, 0.5, 0, 0.5)

    w_present = (w != 0)
    y_present = y * w_present

Instances