Title of Assignment -- k-Means Clustering IT & Computer Science Essay

Student Name
Professor Name
Course
Date
IT Project
Title of Assignment -- k-Means Clustering
1. Set the number of clusters K=3, max run=100, and run the model based on Euclidean distance and show your 1) operators and their connections. 2) parameter settings for the Clustering operator. 3) clustering result using folder view. (0.8 points)
2. What is average within centroid similarity and Davies Bouldin Index under the setting in question 1? Use the following parameter settings for the Performance Operator (0.8 points).
3. The researcher's best friend who is a fragrance mania told the research he would group these perfume brands into the following categories.
Fruity: Burberry, Fendi, Lacoste, Mark Jacobs, Coach, Gucci
Floral: Armani, Chanel, Calvin Klein, Dior, Dolce & Gabbana, Moschino, Prada, Ralph Lauren, YSL
Oceanic: Bvlgari, Chole, Hermes, Issey_Miyake, Versace
With such external criterion, what are the dominant classes in each cluster? (0.6 points)
Calculate purity measure for each cluster (0.6 points)
In the context of cluster analysis, purity is considered an external evaluation of the quality of cluster. It is defined as the percentage of the total number of objects or data points that are to be classified in a unit range.
Purity=1N∑i=1kmaxj|ci∩tj|
Here NN means the total number of objects, kk means the number of clusters and tjtj means the classification which has maximum count for the cluster cici. In order to calculate purity, we will have to follow the following steps:
| T1 | T2 | T3
---------------------
C1 | 0 | 53 | 10
C2 | 0 | 1 | 60
C3 | 0 | 16 | 0
For every cluster cici, we will select the maximum value from ...