Sommelier4U is a company that ships its customers bottles of different types of wine and then has the customers rate the wines as “Like” or

“Dislike.” For each customer, Sommelier4U trains a classification tree based on the characteristics of the wine such as amount of proline and flavonoids in the wine and customer ratings of wines that the customer has tasted. Then, Sommelier4U uses the classification tree to identify new wines that the customer may Like.

Sommelier4U recommends the wines that have a greater than 50% probability of being liked. Neal Jones, a loyal customer, has provided “Like” or “Dislike”

ratings on 178 different wines. Based on these 178 wines, Sommelier4U executed a 10-fold cross-validation experiment to obtain the following pruned classification tree.

a. Construct the confusion matrix resulting from the pruned classification tree on the 178 wines.

b. Sommelier4U plans to apply the pruned classification tree and recommend wine that the tree believes that Neal will like. If Sommelier4U is most concerned that Neal will like the wines it recommends, which performance measure should it focus on? Calculate the value of this measure from the confusion matrix in part a.

c. Consider the wine with the following characteristics: Proline = 820 and Flavonoids = 3.1. Does Sommelier4U believe that Neal will like this wine?

Transcribed Image Text:

Proline < 755 Dislike 97.3%, 108/111 Dislike 66.3%, 118/178 Flavonoids 2.17 Dislike 100%, 8/8 Proline> 755 Like 85.1%, 57/67 Flavonoids > 2.17 Like 96.6%, 57/59