AWS Certified Big Data - Specialty (#46)

A company is using Amazon Machine Learning as part of a medical software application. The application will predict the most likely blood type for a patient based on a variety of other clinical tests that are available when blood type knowledge is unavailable. What is the appropriate model choice and target attribute combination for this problem?

Multi-class classification model with a categorical target attribute.
Regression model with a numeric target attribute.
Binary Classification with a categorical target attribute.
K-Nearest Neighbors model with a multi-class target attribute.