For each of the following features, pick what might be the best encoding for linear classification, among the choices provided. In each part below, you should assume that the dataset has several other features. The questions mention what we're trying to predict, but we're really just asking about the input encoding here. The point of this question is to think about alternatives; there are several options, many not mentioned here.

Car brand, e.g., Chevy, Ford, Toyota, VW, for predicting gas mileage (e.g., low vs. high gas mileage):

(a) An integer valued feature: 1, 2, 3, 4.

(b) Two binary valued features: 00, 01, 10, 11.

(c) 4 unary features (one-hot): 1000, 0100, 0010, 0001.