Return the aggregated prediction across all trees in the RF for each problem. Parameters ---------- predictions : :py:class:`ndarray ` of shape `(n_trees, N)` The array of predictions from each decision tree in the RF for each
(self, predictions)
| 76 | return self._vote(tree_preds) |
| 77 | |
| 78 | def _vote(self, predictions): |
| 79 | """ |
| 80 | Return the aggregated prediction across all trees in the RF for each problem. |
| 81 | |
| 82 | Parameters |
| 83 | ---------- |
| 84 | predictions : :py:class:`ndarray <numpy.ndarray>` of shape `(n_trees, N)` |
| 85 | The array of predictions from each decision tree in the RF for each |
| 86 | of the `N` problems in `X`. |
| 87 | |
| 88 | Returns |
| 89 | ------- |
| 90 | y_pred : :py:class:`ndarray <numpy.ndarray>` of shape `(N,)` |
| 91 | If classifier is True, the class label predicted by the majority of |
| 92 | the decision trees for each problem in `X`. If classifier is False, |
| 93 | the average prediction across decision trees on each problem. |
| 94 | """ |
| 95 | if self.classifier: |
| 96 | out = [np.bincount(x).argmax() for x in predictions.T] |
| 97 | else: |
| 98 | out = [np.mean(x) for x in predictions.T] |
| 99 | return np.array(out) |