Machine Learning and AI Foundations: Clustering and Association

Course details

Unsupervised learning is a type of machine learning where algorithms parse unlabeled data. The focus is not on sorting data into known categories but uncovering hidden patterns. Unsupervised learning plays a big role in modern marketing segmentation, fraud detection, and market basket analysis. This course shows how to use leading machine-learning techniques—cluster analysis, anomaly detection, and association rules—to get accurate, meaningful results from big data.

Instructor Keith McCormick reviews the most common clustering algorithms: hierarchical, k-means, BIRCH, and self-organizing maps (SOM). He uses the same algorithms for anomaly detection, with additional specialized functions available in IBM SPSS Modeler. He closes the course with a review of association rules and sequence detection, and also provides some resources for learning more.

All exercises are demonstrated in IBM SPSS Modeler and IBM SPSS Statistics, but the emphasis is on concepts, not the mechanics of the software.

Learning objectives
What is unsupervised learning?
Cluster and distance-based measures
Hierarchical cluster analysis
K-means cluster analysis
Visualizing and reporting cluster solutions
Cluster methods for categorical variables
Anomaly detection
Association rules
Sequence detection

Concepts

0. Introduction

01 - Welcome
02 - What you should know
03 - Using the exercise files
04 - What is unsupervised machine learning

1. What Is Cluster Analysis

05 - Looking at the data with a 2D scatter plot
06 - Understanding hierarchical cluster analysis
07 - Running hierarchical cluster analysis
08 - Interpreting a dendrogram
09 - Methods for measuring distance
10 - What is k-nearest neighbors

2. K-Means

11 - How does k-means work
12 - Which variables should be used with k-means
13 - Interpreting a box plot
14 - Running a k-means cluster analysis
15 - Interpreting cluster analysis output
16 - What does silhouette mean
17 - Which cases should be used with k-means
18 - Finding optimum value for k - k 3
19 - Finding optimum value for k - k 4
20 - Finding optimum value for k - k 5
21 - What the best solution

3. Visualizing and Reporting Cluster Solutions

22 - Summarizing cluster means in a table
23 - Traffic Light feature in Excel
24 - Line graphs

4. Cluster Methods for Categorical Variables

25 - Relating clusters to categories statistically
26 - Relating clusters to categories visually
27 - Running a multiple correspondence analysis
28 - Interpreting a perceptual map
29 - Using cluster analysis and decision trees together
30 - A BIRCH two-step example
31 - A self organizing map example

5. Anomaly Detection

32 - The k 1 trick
33 - Anomaly detection algorithms
34 - Using SOM for anomaly detection

6. Association Rules and Sequence Detection

35 - Intro to association rules and sequence analysis
36 - Running association rules
37 - Some association rules terminology
38 - Interpreting association rules
39 - Putting association rules to use
40 - Comparing clustering and association rules
41 - Sequence detection

Conclusion

42 - Next steps