CS3481 Fundamentals of Data Science
Assignment 3
Objective
In this assignment, you will study the hierarchical clustering approach introduced in the class
using Python.
Detailed Requirement
We have introduced the hierarchical clustering approach in the class. In this assignment, you will
apply this approach to the Vertebral Column data set from the UCI Machine Learning Repository.
You can perform hierarchical clustering using the method linkage from the module
scipy.cluster.hierarchy.
After performing hierarchical clustering, you could visualize the clustering result in the form of a
dendrogram by using the method dendrogram. You could also study the clustering solution for
a specific number of clusters in the hierarchy by using the method fcluster.
To perform clustering, you should only use the input attributes but not the class label. To improve
the clustering results, you could consider removing outliers from the data set, and applying a
suitable normalization operation to the input attributes.
Assignment Submission
You should submit a report to summarize your work. The following tasks are to be performed:
1. Compare the hierarchical structures generated using single link, complete link and group
average for the Vertebral Column data set. (30%)
2. For some of these hierarchical structures, observe the set of distance values at which cluster
merge occurs, and identify possible patterns from these values. (20%)
3. Select different clustering solutions from the hierarchical structures, and compare the cluster
groupings with the corresponding K-means clustering solutions (using the method KMeans
from the module sklearn.cluster), in terms of the extent to which the clusters can
capture the class structure of the data set. (30%)
4. Select different subsets of attributes from the data sets and re-perform hierarchical clustering.
Compare the resulting hierarchical structures based on the selected attribute subsets with the
original hierarchical structures. (20%)
Please provide a detailed description of the results of the above tasks in your report.
Supplementary Instructions for Assignment 3
To perform hierarchical clustering and K-means clustering in Python, we need to include the
following modules:
from matplotlib import pyplot as plt
from scipy.cluster.hierarchy import dendrogram, linkage, fcluster
from sklearn.cluster import KMeans
Using the Iris data set as an example:
from sklearn import datasets
iris = datasets.load_iris()
X = iris.data
Perform hierarchical clustering using the complete link approach:
Z = linkage(X,'complete')
Visualize the hierarchical clustering result in the form of a dendrogram:
plt.figure(figsize=(25, 10))
dendrogram(Z)
plt.show()
Extract a clustering solution with a specific number of clusters in the hierarchy (The example
below corresponds to the case of three clusters):
kclusters = fcluster(Z, 3, criterion='maxclust')
kclusters
Perform K-means clustering for a given number of clusters (The example below corresponds to
the case of three clusters):
km = KMeans(n_clusters=3)
km.fit(X)
km.labels_