Data Mining
Covers extracting meaningful patterns from large datasets through classification, clustering and association rules, with examples from insurance and finance.
- Semester
- Fall semester
- Level
- Undergraduate
- Credits
- 2+0
- Length
- 16 weeks
Course aims
To enable students to follow the CRISP-DM process, prepare data, choose and apply a suitable data mining method, and interpret results with the right metrics.
Learning outcomes
Students who complete this course will be able to:
- Applies preprocessing steps and assesses data quality.
- Performs classification with decision trees, Naive Bayes, k-NN and ensemble methods.
- Applies and interprets clustering and association rule analyses.
- Compares models using cross-validation and appropriate performance metrics.
Assessment
| Component | Weight |
|---|---|
| Midterm exam (week 8) | 40% |
| Final exam (week 16) | 60% |
Resources
- Jiawei Han, Micheline Kamber, Jian Pei, Data Mining: Concepts and Techniques, Morgan Kaufmann
- Pang-Ning Tan, Michael Steinbach, Anuj Karpatne, Vipin Kumar, Introduction to Data Mining, Pearson
Weekly syllabus
-
1
Week 1 Introduction to data mining and the CRISP-DM process
-
2
Week 2 Data types, attributes and data quality
-
3
Week 3 Preprocessing: cleaning and integration
-
4
Week 4 Transformation, normalisation and reduction
-
5
Week 5 Similarity and distance measures
-
6
Week 6 Classification: decision trees
-
7
Week 7 Classification: Naive Bayes and k-nearest neighbours
-
8
Week 8 Exam Midterm exam
-
9
Week 9 Model evaluation: cross-validation and the confusion matrix
-
10
Week 10 Ensemble methods: bagging, boosting and random forests
-
11
Week 11 Clustering: k-means and k-medoids
-
12
Week 12 Hierarchical and density-based clustering
-
13
Week 13 Association rules: Apriori and FP-Growth
-
14
Week 14 Outlier detection; applications in insurance and finance
-
15
Week 15 Project presentations
-
16
Week 16 Exam Final exam