python - Change Pandas index datatype on MultiIndex -

i have pandas dataframe 2 indexes

                              column1 indexa   indexb                         1001     aaa                        1          bbb                        1          ccc                        1          ddd                        1 

created by

pd.read_sql(sql=sql, index_col=['indexa', 'indexb']) 

mysql reading in indexb unicode , convert string. goal pivot table , have entries in indexb column names. when unicode values following column names:

�  �  �  �  �  �  �  �  � 

when running

pd.read_sql(sql=sql, index_col=['indexa', 'indexb']).unstack().fillna(0) 

edit: comment suggested following:

df = pd.read_sql(sql) df['indexb'] = df['indexb'].astype(str) df = df.set_index(('indexa', 'indexb'), drop=true) 

which nice work around problem (thank you). still nice know if can done during initialization.