problem with predict when multi-class classification

[tmontana]

same for predict, predict_proba and predict_all

preds=automl.predict_proba(X_test)

---

ValueError Traceback (most recent call last)
in
----> 1 preds=automl.predict_proba(X_test)

/anaconda/envs/mlj/lib/python3.8/site-packages/supervised/automl.py in predict_proba(self, X)
312
313 """
--> 314 return self._predict_proba(X)
315
316 def predict_all(self, X):

/anaconda/envs/mlj/lib/python3.8/site-packages/supervised/base_automl.py in _predict_proba(self, X)
878 # If classification task the result is in column 'label'
879 # Need to drop label column.
--> 880 return self._base_predict(X).drop(["label"], axis=1).to_numpy()
881
882 def _predict_all(self, X):

/anaconda/envs/mlj/lib/python3.8/site-packages/supervised/base_automl.py in _base_predict(self, X)
850 target_is_numeric = self._data_info.get("target_is_numeric", False)
851 if target_is_numeric:
--> 852 predictions["label"] = predictions["label"].astype(np.int32)
853 return predictions
854 # Regression

/anaconda/envs/mlj/lib/python3.8/site-packages/pandas/core/generic.py in astype(self, dtype, copy, errors)
5541 else:
5542 # else, only a single dtype is given
-> 5543 new_data = self._mgr.astype(dtype=dtype, copy=copy, errors=errors,)
5544 return self._constructor(new_data).finalize(self, method="astype")
5545

/anaconda/envs/mlj/lib/python3.8/site-packages/pandas/core/internals/managers.py in astype(self, dtype, copy, errors)
593 self, dtype, copy: bool = False, errors: str = "raise"
594 ) -> "BlockManager":
--> 595 return self.apply("astype", dtype=dtype, copy=copy, errors=errors)
596
597 def convert(

/anaconda/envs/mlj/lib/python3.8/site-packages/pandas/core/internals/managers.py in apply(self, f, align_keys, **kwargs)
404 applied = b.apply(f, **kwargs)
405 else:
--> 406 applied = getattr(b, f)(**kwargs)
407 result_blocks = _extend_blocks(applied, result_blocks)
408

/anaconda/envs/mlj/lib/python3.8/site-packages/pandas/core/internals/blocks.py in astype(self, dtype, copy, errors)
592 vals1d = values.ravel()
593 try:
--> 594 values = astype_nansafe(vals1d, dtype, copy=True)
595 except (ValueError, TypeError):
596 # e.g. astype_nansafe can fail on object-dtype of strings

/anaconda/envs/mlj/lib/python3.8/site-packages/pandas/core/dtypes/cast.py in astype_nansafe(arr, dtype, copy, skipna)
965 # work around NumPy brokenness, #1987
966 if np.issubdtype(dtype.type, np.integer):
--> 967 return lib.astype_intsafe(arr.ravel(), dtype).reshape(arr.shape)
968
969 # if we have a datetime/timedelta array of objects

pandas/_libs/lib.pyx in pandas._libs.lib.astype_intsafe()

ValueError: invalid literal for int() with base 10: '1.0'

[pplonski]

Looks like a bug. Thank you for reporting.

@tmontana could you please provide the information about classes that are used in the target? Are they strings, for example "class_1", "class_2", "class_3", ... or other format of the target?

[tmontana]

here's a snapshot for y (they're float64)

0 5.0
1 2.0
2 5.0
3 4.0
4 5.0
...
813296 5.0
813297 5.0
813298 4.0
813299 4.0
813300 1.0
Name: P360d_5_Cats, Length: 813301, dtype: float64

[tmontana]

from 1 to 5

[pplonski]

Got it! Thank you! I will fix it soon.

You can try to cast the target to int and it should fix the problem

df[target] = df[target].astype(int)

[tmontana]

will do thanks. automl._best_model.predict(X_test) also works fyi

[pplonski]

Fixed. It will be in the next release 0.7.3