AttributeError: 'int' object has no attribute 'calc_ders_range'

[fonnesbeck]

Problem:

After running a long cross-validation on a CatBoostRegressor model, CatBoost raises an Attribute error when it tries to fit a model using the optimized hyperparameters:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
_catboost.pyx in _catboost._ObjectiveCalcDersRange()

AttributeError: 'int' object has no attribute 'calc_ders_range'

During handling of the above exception, another exception occurred:

CatBoostError                             Traceback (most recent call last)
<ipython-input-46-bd110940ccc0> in <module>
      2 hyperopt_iterations = 30
      3 
----> 4 model, params = train_best_model(
      5     X, y,
      6     const_params,

<ipython-input-45-34ac36c7d385> in train_best_model(X, y, const_params, max_evals)
     20 
     21     model = CatBoostRegressor(**hyper_params)
---> 22     model.fit(dataset, verbose=False)
     23 
     24     return model, hyper_params

~/anaconda3/envs/draft/lib/python3.8/site-packages/catboost/core.py in fit(self, X, y, cat_features, sample_weight, baseline, use_best_model, eval_set, verbose, logging_level, plot, column_description, verbose_eval, metric_period, silent, early_stopping_rounds, save_snapshot, snapshot_file, snapshot_interval, init_model)
   4657             self._check_is_regressor_loss(params['loss_function'])
   4658 
-> 4659         return self._fit(X, y, cat_features, None, None, sample_weight, None, None, None, None, baseline,
   4660                          use_best_model, eval_set, verbose, logging_level, plot, column_description,
   4661                          verbose_eval, metric_period, silent, early_stopping_rounds,

~/anaconda3/envs/draft/lib/python3.8/site-packages/catboost/core.py in _fit(self, X, y, cat_features, text_features, pairs, sample_weight, group_id, group_weight, subgroup_id, pairs_weight, baseline, use_best_model, eval_set, verbose, logging_level, plot, column_description, verbose_eval, metric_period, silent, early_stopping_rounds, save_snapshot, snapshot_file, snapshot_interval, init_model)
   1736 
   1737         with log_fixup(), plot_wrapper(plot, [_get_train_dir(self.get_params())]):
-> 1738             self._train(
   1739                 train_pool,
   1740                 train_params["eval_sets"],

~/anaconda3/envs/draft/lib/python3.8/site-packages/catboost/core.py in _train(self, train_pool, test_pool, params, allow_clear_pool, init_model)
   1228 
   1229     def _train(self, train_pool, test_pool, params, allow_clear_pool, init_model):
-> 1230         self._object._train(train_pool, test_pool, params, allow_clear_pool, init_model._object if init_model else None)
   1231         self._set_trained_model_attributes()
   1232 

_catboost.pyx in _catboost._CatBoost._train()

_catboost.pyx in _catboost._CatBoost._train()

CatBoostError: catboost/python-package/catboost/helpers.cpp:42: Traceback (most recent call last):
  File "_catboost.pyx", line 1850, in _catboost._ObjectiveCalcDersRange
AttributeError: 'int' object has no attribute 'calc_ders_range'

Has anyone come across this error before?

catboost version: 0.23.2
Operating System: Debian Linux
CPU: AMD

[fonnesbeck]

Some more information: I am using CV to select the best Tweedie variance_power parameter;

'loss_function': hyperopt.hp.choice('loss_function', ['Tweedie:variance_power=1.1',
                                                             'Tweedie:variance_power=1.3',
                                                             'Tweedie:variance_power=1.5',
                                                             'Tweedie:variance_power=1.7'])

which works fine until the best hyperparameters are extracted. The resulting dictionary looks like this:

{'loss_function': 1, 'task_type': 'CPU', 'custom_metric': ['RMSE', 'R2'], 'eval_metric': 'RMSE', 'random_seed': 42}

so, an integer is returned as the best loss function rather than the string of the appropriate function.