Bug: GPBoostError: The argument 'num_neighbors_pred' is discontinued. Use the function 'set_prediction_data' to specify this

Hi, I am currently using GPBoostClassifier inside sklearn's Pipeline.

I noticed in the documentation that GPBoostClassifier's predict method still has a default value of -1 for num_neighbors_pred parameter instead of None even though it is discontinued. This raises an error when calling model.predict(X).

Error is avoided only when GPBoostClassifier.predict(X, num_neighbors_pred = None) is explicitly specified.

Is it possible to rewrite the default value of num_neighbors_pred to None to avoid this error?

GPBoost Version

1.2.1.1

GPBoost Documentation (latest version)

def predict(self, X, raw_score=False, start_iteration=0, num_iteration=None,
                pred_leaf=False, pred_contrib=False,
                group_data_pred=None, group_rand_coef_data_pred=None,
                gp_coords_pred=None, gp_rand_coef_data_pred=None,
                cluster_ids_pred=None, vecchia_pred_type=None,
                num_neighbors_pred=-1, predict_cov_mat=False, predict_var=False, **kwargs):
        """Docstring is inherited from the GPBoostModel."""
        result = self.predict_proba(X=X, raw_score=raw_score, start_iteration=start_iteration, num_iteration=num_iteration,
                                    pred_leaf=pred_leaf, pred_contrib=pred_contrib,
                                    group_data_pred=group_data_pred, group_rand_coef_data_pred=group_rand_coef_data_pred,
                                    gp_coords_pred=gp_coords_pred, gp_rand_coef_data_pred=gp_rand_coef_data_pred,
                                    cluster_ids_pred=cluster_ids_pred, vecchia_pred_type=vecchia_pred_type,
                                    num_neighbors_pred=num_neighbors_pred, predict_cov_mat=predict_cov_mat, predict_var=predict_var,
                                    **kwargs)
        if callable(self._objective) or raw_score or pred_leaf or pred_contrib:
            return result
        else:
            if self._Booster.has_gp_model:
                if self._n_classes > 2:
                    result = result['response_mean']
                else:
                    result = np.vstack((1. - result['response_mean'], result['response_mean'])).transpose()
            class_index = np.argmax(result, axis=1)
            return self._le.inverse_transform(class_index)

Error Message Received

GPBoostError                              Traceback (most recent call last)
File <command-4206059573559381>:1
----> 1 model.predict(x)

File /databricks/python/lib/python3.9/site-packages/sklearn/utils/metaestimators.py:113, in _AvailableIfDescriptor.__get__.<locals>.<lambda>(*args, **kwargs)
    110         raise attr_err
    112     # lambda, but not partial, allows help() to work with update_wrapper
--> 113     out = lambda *args, **kwargs: self.fn(obj, *args, **kwargs)  # noqa
    114 else:
    116     def fn(*args, **kwargs):

File /databricks/python/lib/python3.9/site-packages/sklearn/pipeline.py:470, in Pipeline.predict(self, X, **predict_params)
    468 for _, name, transform in self._iter(with_final=False):
    469     Xt = transform.transform(Xt)
--> 470 return self.steps[-1][1].predict(Xt, **predict_params)

File <command-1775996920262159>:21, in predict(self, X, y)

File /local_disk0/.ephemeral_nfs/envs/pythonEnv-3330d11d-731b-439d-b8b9-5adea6365f78/lib/python3.9/site-packages/gpboost/sklearn.py:941, in GPBoostClassifier.predict(self, X, raw_score, start_iteration, num_iteration, pred_leaf, pred_contrib, group_data_pred, group_rand_coef_data_pred, gp_coords_pred, gp_rand_coef_data_pred, cluster_ids_pred, vecchia_pred_type, num_neighbors_pred, predict_cov_mat, predict_var, **kwargs)
    934 def predict(self, X, raw_score=False, start_iteration=0, num_iteration=None,
    935             pred_leaf=False, pred_contrib=False,
    936             group_data_pred=None, group_rand_coef_data_pred=None,
    937             gp_coords_pred=None, gp_rand_coef_data_pred=None,
    938             cluster_ids_pred=None, vecchia_pred_type=None,
    939             num_neighbors_pred=-1, predict_cov_mat=False, predict_var=False, **kwargs):
    940     """Docstring is inherited from the GPBoostModel."""
--> 941     result = self.predict_proba(X=X, raw_score=raw_score, start_iteration=start_iteration, num_iteration=num_iteration,
    942                                 pred_leaf=pred_leaf, pred_contrib=pred_contrib,
    943                                 group_data_pred=group_data_pred, group_rand_coef_data_pred=group_rand_coef_data_pred,
    944                                 gp_coords_pred=gp_coords_pred, gp_rand_coef_data_pred=gp_rand_coef_data_pred,
    945                                 cluster_ids_pred=cluster_ids_pred, vecchia_pred_type=vecchia_pred_type,
    946                                 num_neighbors_pred=num_neighbors_pred, predict_cov_mat=predict_cov_mat, predict_var=predict_var,
    947                                 **kwargs)
    948     if callable(self._objective) or raw_score or pred_leaf or pred_contrib:
    949         return result

File /local_disk0/.ephemeral_nfs/envs/pythonEnv-3330d11d-731b-439d-b8b9-5adea6365f78/lib/python3.9/site-packages/gpboost/sklearn.py:1033, in GPBoostClassifier.predict_proba(self, X, raw_score, start_iteration, num_iteration, pred_leaf, pred_contrib, group_data_pred, group_rand_coef_data_pred, gp_coords_pred, gp_rand_coef_data_pred, cluster_ids_pred, vecchia_pred_type, num_neighbors_pred, predict_cov_mat, predict_var, **kwargs)
    961 def predict_proba(self, X, raw_score=False, start_iteration=0, num_iteration=None,
    962                   pred_leaf=False, pred_contrib=False,
    963                   group_data_pred=None, group_rand_coef_data_pred=None,
    964                   gp_coords_pred=None, gp_rand_coef_data_pred=None,
    965                   cluster_ids_pred=None, vecchia_pred_type=None,
    966                   num_neighbors_pred=None, predict_cov_mat=False, predict_var=False, **kwargs):
    967     """Return the predicted probability for each class for each sample.
    968 
    969     Parameters
   (...)
   1031         If ``pred_contrib=True``, the feature contributions for each sample.
   1032     """
-> 1033     result = super().predict(X=X, raw_score=raw_score, start_iteration=start_iteration, num_iteration=num_iteration,
   1034                              pred_leaf=pred_leaf, pred_contrib=pred_contrib,
   1035                              group_data_pred=group_data_pred, group_rand_coef_data_pred=group_rand_coef_data_pred,
   1036                              gp_coords_pred=gp_coords_pred, gp_rand_coef_data_pred=gp_rand_coef_data_pred,
   1037                              cluster_ids_pred=cluster_ids_pred, vecchia_pred_type=vecchia_pred_type,
   1038                              num_neighbors_pred=num_neighbors_pred, predict_cov_mat=predict_cov_mat,
   1039                              predict_var=predict_var, **kwargs)
   1040     if callable(self._objective) and not (raw_score or pred_leaf or pred_contrib):
   1041         _log_warning("Cannot compute class probabilities or labels "
   1042                      "due to the usage of customized objective function.\n"
   1043                      "Returning raw scores instead.")

File /local_disk0/.ephemeral_nfs/envs/pythonEnv-3330d11d-731b-439d-b8b9-5adea6365f78/lib/python3.9/site-packages/gpboost/sklearn.py:752, in GPBoostModel.predict(self, X, raw_score, start_iteration, num_iteration, pred_leaf, pred_contrib, group_data_pred, group_rand_coef_data_pred, gp_coords_pred, gp_rand_coef_data_pred, cluster_ids_pred, vecchia_pred_type, num_neighbors_pred, predict_cov_mat, predict_var, **kwargs)
    747 if self._n_features != n_features:
    748     raise ValueError("Number of features of the model must "
    749                      "match the input. Model n_features_ is %s and "
    750                      "input n_features is %s "
    751                      % (self._n_features, n_features))
--> 752 return self._Booster.predict(X, pred_latent=raw_score, start_iteration=start_iteration, num_iteration=num_iteration,
    753                              pred_leaf=pred_leaf, pred_contrib=pred_contrib,
    754                              group_data_pred=group_data_pred, group_rand_coef_data_pred=group_rand_coef_data_pred,
    755                              gp_coords_pred=gp_coords_pred, gp_rand_coef_data_pred=gp_rand_coef_data_pred,
    756                              cluster_ids_pred=cluster_ids_pred, vecchia_pred_type=vecchia_pred_type,
    757                              num_neighbors_pred=num_neighbors_pred, predict_cov_mat=predict_cov_mat,
    758                              predict_var=predict_var, **kwargs)

File /local_disk0/.ephemeral_nfs/envs/pythonEnv-3330d11d-731b-439d-b8b9-5adea6365f78/lib/python3.9/site-packages/gpboost/basic.py:3516, in Booster.predict(self, data, start_iteration, num_iteration, pred_latent, pred_leaf, pred_contrib, data_has_header, is_reshape, group_data_pred, group_rand_coef_data_pred, gp_coords_pred, gp_rand_coef_data_pred, cluster_ids_pred, predict_cov_mat, predict_var, cov_pars, ignore_gp_model, raw_score, vecchia_pred_type, num_neighbors_pred, **kwargs)
   3513     raise GPBoostError("The argument 'vecchia_pred_type' is discontinued. " 
   3514                        "Use the function 'set_prediction_data' to specify this")
   3515 if num_neighbors_pred is not None:
-> 3516     raise GPBoostError("The argument 'num_neighbors_pred' is discontinued. " 
   3517                        "Use the function 'set_prediction_data' to specify this")
   3518 predictor = self._to_predictor(deepcopy(kwargs))
   3519 if num_iteration is None:

GPBoostError: The argument 'num_neighbors_pred' is discontinued. Use the function 'set_prediction_data' to specify this

fabsig / GPBoost