I am running machine learning using name features to predict Y (binary 0 and 1 labels).
Using the name entity (eg: John Carter), I derive into 4 name substring features (1: first name = "John", 2: last name = "Carter, 3: first letter of last name = "c", and 4: last letter of last name = "r").
Obviously "John Carter" is just one record, and other names "Jennifer Green" and "Kyle Ferguson" will give different entries.
Before running the ML algorithm, I needed to transform the dataset into a sparse matrix via vectorization, the "new" columns will be based on the substring features found from actual cases in my data. For example for John Carter, the columns "first-name=John" and "last-name=Carter" will have the value 1, while Jennifer Green will have a value of 0 in these columns, and vice versa.
So is it correct to say that my model contains 4 name features, or should I say it contains, say, tens of thousand of features?
