I am new to ML and this forum so please pardon my ignorance. I am working on the housing dataset from Kaggle and was looking at the OrdinalEncoder.
My question is shouldn’t there be an option to provide OrdinalEncoder with curated ordering in case the natural ordering does not make sense? For example for categories as given below
| Ex | Excellent
| Gd | Good
| TA | Typical/Average
| Fa | Fair
| Po | Poor
It seems there is an inherent ordering to these category values(alphabetical) which is not the same as semantic ordering(EX, Gc, TA, Fa, Po). My impression is this information of order among the categories is useful and can be used by some ML Models like Linear Regression? What is your opinion