Discriminant analysis, k-Nearest Neighbors, and Naïve Bayes are all analytic methods used to __________.

• A. Classify only
• B. Estimate numeric values
• C. Predict and categorize
• D. Reduce dimensionality

Answer: (C)

These methods are supervised learning tools that can take labeled data and help you decide what category a new observation belongs to or estimate a numeric value for it. Discriminant analysis builds rules to separate classes and assigns a class label (and often class probabilities) to new data. k-Nearest Neighbors looks at nearby labeled examples to decide the new instance’s category, and it can also be used for regression to predict a numeric outcome. Naïve Bayes combines feature evidence to compute the probability of each class and typically outputs the most likely category. Because each method can yield a class decision and, in applicable setups, a numeric prediction, “predict and categorize” best captures their common use.