Question

Business Analytics, Assignment on Clustering As part of the quarterly reviews, the manager of a retail store analyzes the quality of customer service based on the periodic customer satisfaction ratings (on a scale of 1 to 10 with 1 = Poor and 10 = Excellent). To understand the level of service quality, which includes the waiting times of the customers in the checkout section, he collected data on 100 customers who visited the store; see the attached Excel file: ServiceQuality. 1. Using Data Mining > Cluster, apply K-Means Clustering with the following Selected Variables: Wait Time (min), Purchase Amount ($), Customer Age, and Customer Satisfaction Rating. In Step 2 of the k-Means Clustering procedure, normalize (standardize) input data, assume k = 5 clusters, 50 iterations, and Fixed start with the default Centroid Initialization seed of 12345. In Step 3, select the checkboxes “Show data summary” and “Show distances from each cluster center”. a. What is the most homogenous cluster? What is the number of customers in this cluster? For this cluster, what is the average standardized Euclidean distance between its observations and its centroid (center)? What is the centroid of this cluster (expressed in standardized data)? Using the cluster centroids, how would you characterize the customers in the most homogenous cluster in comparison with the customers in the remaining clusters? b. Which two clusters are most distinct and why? Using their centroids, how would you compare the customers of the two clusters? 2. Using Data Mining > Cluster, apply Hierarchical Clustering with the following Selected Variables: Wait Time (min), Purchase Amount ($), Customer Age, and Customer Satisfaction Rating. In Step 2 of the Hierarchical Clustering procedure, normalize (standardize) input data and apply Ward’s clustering method, while in Step 3, select the checkboxes “Show dendrogram”, “Show Cluster Membership”, and assume k = 5 clusters. a. Show the obtained dendrogram. b. What are the sizes of the created clusters? c. What are the centroids of the created clusters expressed in original data. Hint: To answer Questions 2b and 2c, in the worksheet HC_Clusters, you may first add the four columns with original data, and sort the column Cluster by clicking Data in the Ribbon and selecting . Then you can easily find the cluster sizes, and the four variable averages characterizing each cluster. 3. Based on your findings, what reasons do you see for low customer satisfaction ratings? Provide some recommendations for improving customer satisfaction. Write a managerial report in MS Word; do not attach any separate Data Mining and/or Excel outputs. Instead paste into your report the Data Mining outputs with answers to Questions in Tasks 1 and 2a, and Excel results related to Tasks 2b and 2c. Showing these outputs/results is crucial because without them I will not be able to verify your answers. Note 1. The use of other software is unacceptable! Note 2. To better prepare for the assignment, study the clustering example KTC in the slides, and the questions on Exhibit 4.4 in Test Bank.

Answer 1

within cluster sum of squares by cluster: [1] 11.374864 2.635716 4.056018 2.268045 3.209352 (between-SS / total-SS 91.0 %) - Above answer your question like, V4 in the image is nothing but the satisfaction rate.

0.0 0.2 0.4 0.6 0.8 o O o Wait Time..min o O O Oo Purchase Amount. … O o 0 O O 0 O Customer Age 0.0 0.2 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8

Above graph, will depict you variables with less density of cluster. Purchase amount formed a nice dense cluster which with we can analyze and get more insights.

2) 0.0 02 0 06 0.10 Purchase Amount Customer Age V4 0.0 02 04 06 8 10 Above image has 3 clusters where k = 3 and cluster is formed beautifully between the variable customer Age and purchase amount. In the previous case, when cluster k 5, grouping is that much good because we didn't form an optimal cluster group.

Rest of the task can be answered by following dendrogram build based on Ward method When K =3, Cluster Dendrogram

hclust (, "ward D") when k =5, Cluster Dendrogram 5 工寸