np.select dtype error when assigning feature group labels

[KiraLiiu]

When running the preprocessing pipeline (specifically in preprocess.py), the following error occurs:

Error processing configuration: Choicelist and default value do not have a common dtype: The DType <class 'numpy.dtypes._PyLongDType'> could not be promoted by <class 'numpy.dtypes.StrDType'>. This means that no common DType exists for the given inputs. For example they cannot be stored in a single array unless the dtype is `object`.

This happens at the line:

choicelist = feature_categories.keys()
df = df.assign(group=np.select(condlist, choicelist))

How to Fix

• Explicitly convert feature_categories.keys() to a list of strings using list().
• Explicitly set the default parameter in np.select to a string value (e.g., 'other'), ensuring all possible outputs are of the same type.

Solution

Change this code:

choicelist = feature_categories.keys()
df = df.assign(group=np.select(condlist, choicelist))

to:

choicelist = list(feature_categories.keys())
df = df.assign(group=np.select(condlist, choicelist, default='other'))

This ensures that both choicelist and default are of the same type (str), resolving the dtype error.