Printable Handouts

PDF

Navigable Slide Index

1. Introduction
2. The problem
3. Simplicity first!
4. Agenda
5. One attribute does all the work
6. Example
7. Complications: missing values
8. Complications: overfitting
9. Evaluating the result
10. First description of the method
11. Agenda: statistical modeling
12. Statistical modeling
13. Bayes's rule
14. Weather data: probabilities
15. Weather data: results
16. Complications: zero frequencies, missing values
17. Numeric attributes
18. Numeric attributes: example
19. "Nave" statistical model
20. Agenda: constructing decision trees
21. Constructing decision trees
22. Which attribute to select? (1)
23. Which is the best attribute?
24. Which attribute to select? (2)
25. Continuing to split
26. Complications: highly-branching attributes
27. Complications: overfitting and the need to prune (1)
28. Complications: overfitting and the need to prune (2)
29. Complications: missing values, numeric attributes
30. Constructing decision trees: summary
31. Agenda: constructing rules
32. Constructing rules
33. Generating a rule
34. Rules vs. trees
35. Constructing rules: general algorithm
36. More about rules
37. Association rules
38. Constructing association rules
39. Association rules: example
40. Association rules: discussion

Topics Covered

• Simplicity first
• One attribute does all the work
• Complications: missing values
• Overfitting
• Evaluating the result
• Description of the method
• Statistical modeling
• Bayes's rule
• Weather data: probabilities
• Zero frequencies, missing values
• Numeric attributes
• "Naive" statistical model
• Constructing decision trees
• Continuing to split
• Highly-branching attributes
• Overfitting and the need to prune
• Constructing rules
• Rules versus trees
• General algorithm
• Association rules

Links

Series:

• Data Mining

Categories:

• Technology & Operations

Talk Citation

Publication History

• Published on June 17, 2008
• Archived on March 17, 2022