- Question 3 6 Points Sonia Is A Manager At A Health Insurance Company She Needs To Identify Her Company S Customers Wh 1 (84.2 KiB) Viewed 31 times
- Question 3 6 Points Sonia Is A Manager At A Health Insurance Company She Needs To Identify Her Company S Customers Wh 2 (58.48 KiB) Viewed 31 times
Question 3A (3 Points): Which version of the analysis (i.e.,
k=3, k=4, or k=5) is better in meeting
Sonia’s objectives? Briefly describe why you selected this version
(1-2 sentences only).
Question 3B (3 Points): Consider the version of the analysis you
selected in Question 3A above.
Which cluster of customers from the selected version of analysis
should be enrolled in the pilot
program? Explain why you selected this particular cluster (1-2
sentences only).
QUESTION 3 (6 Points) Sonia is a manager at a health insurance company. She needs to identify her company's customers who may be most at risk of developing coronary heart disease. Customers identified at risk would then be invited to enroll in a pilot health management program to help them avoid heart disease through dietary and exercise initiatives. Sonia, however, likes to limit the number of participants in the pilot program in order to minimize cost. Also, another of Sonia's objectives is to enroll those customers in the pilot program who are most at risk of coronary heart disease. Sonia has customer data which contains 547 records and features (columns) such as gender, weight and cholesterol. High weight and high cholesterol are generally associated with coronary heart disease. Sonia has conducted k-means cluster analysis on the customer data three times, with k = 3, k = 4, and k = 5. Results of the three versions of the cluster analysis are copied below for your reference. K= 3 Analysis Results Cluster 0: 191 items Cluster 1: 185 items Cluster 2: 171 items Total number of items: 547 Attribute Weight cluster_0 110.461 125.979 cluster_1 141.995 173.249 cluster_2 182.263 217.041 Cholesterol Gender 0.550 0.411 0.585 K= 4 Analysis Results Attribute cluster_0 cluster_2 cluster_1 106.850 cluster_3 184.318 127.726 152.093 Weight Cholesterol 154.385 119.536 185.907 218.916 Gender 0.459 0.543 0.441 0.591
Cluster 2: 118 items Cluster 3: 154 items Total number of items: 547 K= 5 Analysis Results Cluster 0: 107 items Cluster 1: 101 items Cluster 2: 104 items Cluster 3: 110 items Cluster 4: 125 items Total number of items: 547 Attribute cluster_0 cluster_1 cluster_2 cluster_3 cluster_4 Weight 139.028 160.525 120.260 104.618 187.440 Cholesterol 168.551 196.861 142.827 115.864 221.680 Gender 0.430 0.535 0.490 0.582 0.528 Question 3A (3 Points): Which version of the analysis (i.e., k=3, k=4, or k=5) is better in meeting Sonia's objectives? Briefly describe why you selected this version (1-2 sentences only).