TypeError using the Box-Cox method

[danielavornic]

Code

from imperio import BoxCoxTransformer

boxcox = BoxCoxTransformer()
boxcox.fit(X_train, y_train)
X_transformed = boxcox.transform(X_test)

Data Set

Heart Failure Prediction Dataset

Expected Behaviour

The Box-Cox feature selection method should transform the test data using the BoxCoxTransformer without any errors.

Actual Behaviour

When executing the transform method of the BoxCoxTransformer, the following error is raised:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[84], line 10
      8 boxcox = BoxCoxTransformer()
      9 boxcox.fit(X_train, y_train)
---> 10 X_transformed = boxcox.transform(X_test)

File ~/.local/lib/python3.10/site-packages/imperio/BoxCox.py:77, in BoxCoxTransformer.transform(self, X, **fit_params)
     75             return X_copy
     76         else:
---> 77             return np.array([self.__apply_change(xi) for xi in X])

File ~/.local/lib/python3.10/site-packages/imperio/BoxCox.py:77, in <listcomp>(.0)
     75             return X_copy
     76         else:
---> 77             return np.array([self.__apply_change(xi) for xi in X])

File ~/.local/lib/python3.10/site-packages/imperio/BoxCox.py:47, in BoxCoxTransformer.__apply_change(self, value)
     39 def __apply_change(self, value : float) -> float:
     40     '''
     41         The functions that transforms a scalar value.
     42     :param value: float
   (...)
     45         The transformed value by BoxCoxTransformer.
     46     '''
---> 47     return np.log(value) if self.l == 0 else (self.l ** (-1)) * (value ** self.l - 1)

TypeError: unsupported operand type(s) for ** or pow(): 'str' and 'float'

Additional Information

To provide more context and assist with debugging, this is the Jupyter Notebook where I encountered this error and attempted to resolve it.