Fix unittests by defaulting n_classes=2

from scipy.stats import pearsonr

def to_one_hot(y, axis=-1):

def to_one_hot(y, axis=-1, n_classes=2):

  """Transforms label vector into one-hot encoding.

  Turns y into vector of shape [n_samples, 2] (assuming binary labels).

    A vector of shape [n_samples, 1]

  """

  n_samples = np.shape(y)[0]

  n_classes = np.max(y, axis=axis)

  y_hot = np.zeros((n_samples, n_classes))

  y_hot[np.arange(n_samples), y.astype(np.int64)] = 1

  return y_hot

    if self.features is not None:

      feed_dict[self.features[0].out_tensor] = X_b

    return feed_dict

  def predict_on_batch(self, X, tf_initialized=False):

    # sample x task

    # Class is implied by the value of task [0,1]

    prediction = super(MultiTaskTensorGraph, self).predict_on_batch(X)

    prediction = np.transpose(from_one_hot(prediction, axis=2))

    return prediction

  def predict_proba_on_batch(self, X, tf_initialized=False):

    prediction = super(MultiTaskTensorGraph, self).predict_on_batch(X)

    # sample x task x class

    prediction1 = np.transpose(prediction, axes=[1, 0, 2])

    return prediction1

    np.random.seed(123)

    ids = np.arange(n_samples)

    X = np.random.randint(2, size=(n_samples, n_features))

    y = np.zeros((n_samples, n_tasks))

    y = np.ones((n_samples, n_tasks))

    w = np.ones((n_samples, n_tasks))

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

    IRV_transformer = dc.trans.IRVTransformer(5, n_tasks, dataset)