Saved Model Cannot Be Used for Prediction When Saved with Model_Only=True

[maverick-scientist]

Hi Team,
I saved a lot of models with model_only set to true. I reopened my notebook, cleansed the data and executed the "setup" method. Then loaded all the saved models and tried to run predict method only to receive the below error. I was under the impression that if I saved a model with model_only=True, I need to run the setup and the model can be used for everything without any hassle, Which doesn't seem to be the case.

ERROR

---

ValueError Traceback (most recent call last)
Input In [71], in <cell line: 1>()
----> 1 unseen_predictions_tuned_model = predict_model(tuned_model, data=data_unseen)
2 unseen_predictions_tuned_model.head()

File ~\AppData\Roaming\Python\Python38\site-packages\pycaret\classification.py:2126, in predict_model(estimator, data, probability_threshold, encoded_labels, raw_score, drift_report, round, verbose, drift_kwargs)
2047 def predict_model(
2048 estimator,
2049 data: Optional[pd.DataFrame] = None,
(...)
2056 drift_kwargs: Optional[dict] = None,
2057 ) -> pd.DataFrame:
2059 """
2060 This function predicts Label and Score (probability of predicted
2061 class) using a trained model. When data is None, it predicts label and
(...)
2123
2124 """
-> 2126 return pycaret.internal.tabular.predict_model(
2127 estimator=estimator,
2128 data=data,
2129 probability_threshold=probability_threshold,
2130 encoded_labels=encoded_labels,
2131 raw_score=raw_score,
2132 drift_report=drift_report,
2133 round=round,
2134 verbose=verbose,
2135 ml_usecase=MLUsecase.CLASSIFICATION,
2136 drift_kwargs=drift_kwargs,
2137 )

File ~\AppData\Roaming\Python\Python38\site-packages\pycaret\internal\tabular.py:9116, in predict_model(estimator, data, probability_threshold, encoded_labels, drift_report, raw_score, round, verbose, ml_usecase, display, drift_kwargs)
9113 probability_threshold = estimator.probability_threshold
9114 estimator = get_estimator_from_meta_estimator(estimator)
-> 9116 pred = np.nan_to_num(estimator.predict(X_test_))
9118 try:
9119 score = estimator.predict_proba(X_test_)

File ~\AppData\Roaming\Python\Python38\site-packages\sklearn\utils\metaestimators.py:119, in _IffHasAttrDescriptor.get..(*args, **kwargs)
116 attrgetter(self.delegate_names[-1])(obj)
118 # lambda, but not partial, allows help() to work with update_wrapper
--> 119 out = lambda *args, **kwargs: self.fn(obj, *args, **kwargs)
120 # update the docstring of the returned function
121 update_wrapper(out, self.fn)

File ~\AppData\Roaming\Python\Python38\site-packages\sklearn\pipeline.py:408, in Pipeline.predict(self, X, **predict_params)
406 for _, name, transform in self._iter(with_final=False):
407 Xt = transform.transform(Xt)
--> 408 return self.steps[-1][-1].predict(Xt, **predict_params)

File ~\AppData\Roaming\Python\Python38\site-packages\pycaret\internal\meta_estimators.py:151, in CustomProbabilityThresholdClassifier.predict(self, X, **predict_params)
149 if not hasattr(self.classifier_, "predict_proba"):
150 return self.classifier_.predict(X, **predict_params)
--> 151 pred = self.classifier_.predict_proba(X, **predict_params)
152 if pred.shape[1] > 2:
153 raise ValueError(
154 f"{self.class.name} can only be used for binary classification."
155 )

File ~\AppData\Roaming\Python\Python38\site-packages\sklearn\ensemble_forest.py:673, in ForestClassifier.predict_proba(self, X)
671 check_is_fitted(self)
672 # Check data
--> 673 X = self._validate_X_predict(X)
675 # Assign chunk of trees to jobs
676 n_jobs, _, _ = _partition_estimators(self.n_estimators, self.n_jobs)

File ~\AppData\Roaming\Python\Python38\site-packages\sklearn\ensemble_forest.py:421, in BaseForest.validate_X_predict(self, X)
417 """
418 Validate X whenever one tries to predict, apply, predict_proba."""
419 check_is_fitted(self)
--> 421 return self.estimators[0]._validate_X_predict(X, check_input=True)

File ~\AppData\Roaming\Python\Python38\site-packages\sklearn\tree_classes.py:396, in BaseDecisionTree.validate_X_predict(self, X, check_input)
394 n_features = X.shape[1]
395 if self.n_features != n_features:
--> 396 raise ValueError("Number of features of the model must "
397 "match the input. Model n_features is %s and "
398 "input n_features is %s "
399 % (self.n_features_, n_features))
401 return X

ValueError: Number of features of the model must match the input. Model n_features is 233 and input n_features is 235.

There has been no change in the source data and all the steps executed were same.

Please advise.

Thanks in advance,
Abhinav