Mixed real and boolean features

From Knowledge Discovery

Jump to: navigation, search

Given a data set which has e.g. X1: A,B X2: A, B,C X3: real you need to generate 6 features, each of which is binary

X1A X1B X2A X2B X2C X3


Then a given observations X1=A, X2 = A, X3 = 2.2 would give the vector 1 0 1 0 0 2.2

Actually, you could code save a bit of redundancy and code the features using

X1A X2A X2B X3

and - of course - the constant term.

Personal tools