“Specifying the columns using strings is only supported for pandas DataFrames” while using pipeline.fit()

Question

Code: I am using a different columns for my actual application but have created numeric_features and categorical feature in the same way as defined in the sklearn documentat