Suppose you are given a data set consisting of nominal attributes, such as color, which takes values such as red, blue, green etc. can you use this data set directly to train an svm? if not, how will you transform a nominal attribute into a representation that can be used to train an svm?

No, you can´t.

Explanation:

An SVM uses a binary system. In this case the data consists of nominal attributes which can´t be used directly into the SVM.

However, in order to transform this normal attribute into the binary system is necessary to order the information in two sequences, in which the colors are based into two information facts that will give the system the representation that needs.

a hotel that serge might be staying in

answerinformation systems literacy;