How to use the lightfm.data._IncrementalCOOMatrix function in lightfm

def build(self, data):

        features = _IncrementalCOOMatrix(self.features_shape(), np.float32)

        if self._identity_features:
            for (_id, idx) in self._id_mapping.items():
                features.append(idx, self._feature_mapping[_id], 1.0)

        for datum in data:
            for (entity_idx, feature_idx, weight) in self._process_features(datum):
                features.append(entity_idx, feature_idx, weight)

        features = features.tocoo().tocsr()

        if self._normalize:
            if np.any(features.getnnz(1) == 0):
                raise ValueError(
                    "Cannot normalize feature matrix: some rows have zero norm. "
                    "Ensure that features were provided for all entries."

data: iterable of (user_id, item_id) or (user_id, item_id, weight)
            An iterable of interactions. The user and item ids will be
            translated to internal model indices using the mappings
            constructed during the fit call. If weights are not provided
            they will be assumed to be 1.0.

        Returns
        -------

        (interactions, weights): COO matrix, COO matrix
            Two COO matrices: the interactions matrix
            and the corresponding weights matrix.
        """

        interactions = _IncrementalCOOMatrix(self.interactions_shape(), np.int32)
        weights = _IncrementalCOOMatrix(self.interactions_shape(), np.float32)

        for datum in data:
            user_idx, item_idx, weight = self._unpack_datum(datum)

            interactions.append(user_idx, item_idx, 1)
            weights.append(user_idx, item_idx, weight)

        return (interactions.tocoo(), weights.tocoo())

data: iterable of (user_id, item_id) or (user_id, item_id, weight)
            An iterable of interactions. The user and item ids will be
            translated to internal model indices using the mappings
            constructed during the fit call. If weights are not provided
            they will be assumed to be 1.0.

        Returns
        -------

        (interactions, weights): COO matrix, COO matrix
            Two COO matrices: the interactions matrix
            and the corresponding weights matrix.
        """

        interactions = _IncrementalCOOMatrix(self.interactions_shape(), np.int32)
        weights = _IncrementalCOOMatrix(self.interactions_shape(), np.float32)

        for datum in data:
            user_idx, item_idx, weight = self._unpack_datum(datum)

            interactions.append(user_idx, item_idx, 1)
            weights.append(user_idx, item_idx, weight)

        return (interactions.tocoo(), weights.tocoo())