from sklearn.preprocessing import OneHotEncoder
import numpy as np
from sklearn.preprocessing import OneHotEncoder

# I know all my categories in advance and there will be no other
# Using only one sample to call fit since I need to call at some point

list_of_bounds = [["as","mmas","eas","ras","acs"], ["1", "2"]]

one_sample = ["as", "1"]

o_h_enc = OneHotEncoder(categories=list_of_bounds)
o_h_enc.fit(np.array(one_sample).reshape(1, -1) )


print(o_h_enc.categories_)
# [array(['as', 'mm', 'ea', 'ra', 'ac'], dtype='<U2'), array(['1', '2'], dtype='<U2')]
# The first category should be dtype object and contain the full names i.e.
# [array(["as","mmas","eas","ras","acs"], dtype='object'), array(['1', '2'], dtype='<U2')]

o_h_enc.transform(np.array(["mmas", "1"]).reshape((1, -1)))