How to use the alibi.utils.mapping.ord_to_num function in alibi

if self.enc_model:
            enc_data = self.enc.predict(train_data)
            self.class_proto = {}  # type: dict
            self.class_enc = {}  # type: dict
            for i in range(self.classes):
                idx = np.where(preds == i)[0]
                self.class_proto[i] = np.expand_dims(np.mean(enc_data[idx], axis=0), axis=0)
                self.class_enc[i] = enc_data[idx]
        elif self.use_kdtree:
            logger.warning('No encoder specified. Using k-d trees to represent class prototypes.')
            if trustscore_kwargs is not None:
                ts = TrustScore(**trustscore_kwargs)
            else:
                ts = TrustScore()
            if self.is_cat:  # map categorical to numerical data
                train_data = ord_to_num(train_data_ord, self.d_abs)
            ts.fit(train_data, preds, classes=self.classes)
            self.kdtrees = ts.kdtrees
            self.X_by_class = ts.X_kdtree

x = np.argmax(x)
            return x != y

        # define target classes for prototype if not specified yet
        if target_class is None:
            target_class = list(range(self.classes))
            target_class.remove(np.argmax(Y, axis=1))
            if verbose:
                print('Predicted class: {}'.format(np.argmax(Y, axis=1)))
                print('Target classes: {}'.format(target_class))

        if self.is_cat and self.ohe:  # map categorical to numerical data
            X_ord = ohe_to_ord(X, self.cat_vars)[0]
            X_num = ord_to_num(X_ord, self.d_abs)
        elif self.is_cat:
            X_num = ord_to_num(X, self.d_abs)
        else:
            X_num = X

        # find closest prototype in the target class list
        dist_proto = {}
        if self.enc_model:

            X_enc = self.enc.predict(X)
            class_dict = self.class_proto if k is None else self.class_enc

            for c, v in class_dict.items():
                if c not in target_class:
                    continue
                if k is None:
                    dist_proto[c] = np.linalg.norm(X_enc - v)
                elif k is not None:

x = np.copy(x)
                x[y] += self.kappa
                x = np.argmax(x)
            return x != y

        # define target classes for prototype if not specified yet
        if target_class is None:
            target_class = list(range(self.classes))
            target_class.remove(np.argmax(Y, axis=1))
            if verbose:
                print('Predicted class: {}'.format(np.argmax(Y, axis=1)))
                print('Target classes: {}'.format(target_class))

        if self.is_cat and self.ohe:  # map categorical to numerical data
            X_ord = ohe_to_ord(X, self.cat_vars)[0]
            X_num = ord_to_num(X_ord, self.d_abs)
        elif self.is_cat:
            X_num = ord_to_num(X, self.d_abs)
        else:
            X_num = X

        # find closest prototype in the target class list
        dist_proto = {}
        if self.enc_model:

            X_enc = self.enc.predict(X)
            class_dict = self.class_proto if k is None else self.class_enc

            for c, v in class_dict.items():
                if c not in target_class:
                    continue
                if k is None: