Intelligent data analysis

DV713A, Advanced level, 7.5 ECTS (official page in Swedish here)
DV911F, Research level, 7.5 ECTS
Examiner and responsible: Henrik Boström

Course goals

The student should after completing the course be able to:

• develop decision support solutions using techniques from the field of intelligent data analysis
• choose data representation and method for a particular problem
• perform analyses and report results from these
• critically review decision support solutions

The course consists of 9 lectures à two hours and a mandatory seminar (see schedule for dates and location).

Lecture	Slides	Reading	Links
1. Course overview,     The knowledge discovery process	L1(Updated: Nov. 11, 2008)	Ch. 1, 2	CRISP-DM
2. Evaluation	L2 (Updated: Nov. 20, 2008)	Ch. 5
3. Decision trees	L3 (No update required)	Ch. 3.2, 4.3, 6.1
4. Rules	L4 (Updated: Nov. 27, 2008)	Ch. 3.3, 4.4, 6.2
5. Linear models and instance-based learning	L5 (Updated: Nov. 27, 2008)	Ch. 3.8, 4.6, 4.7, 6.3, 6.4
6. Combining multiple models	L6 (Updated: Dec. 4, 2008)	Ch. 7.5
7. Bayesian networks	L7 (Updated: Dec. 4, 2008)	Ch. 6.7
8. Developing and deploying predictive models*	L8 (Updated: Dec. 16, 2008)
9. Association rules and clustering	L9 (Updated: Dec. 18, 2008)	Ch. 3.4, 3.9, 4.7, 4.8, 6.4, 6.6

*Guest lecturer Ulf Norinder, AstraZeneca R&D Södertälje, Sweden

NOTE: If not marked as updated, the slides above are from 2007. They will be updated during the course.

Literature

• Course book:
  Ian H. Witten and Eibe Frank, Data Mining: Practical Machine Learning Tools and Techniques (Second Edition), Morgan Kaufmann, 2005 
• Selected papers and reports (see links in above table).

Examination for DV713A

• Written examination, 5 ECTS (see schedule)
  Two parts: the analysis process and evaluation (2 ECTS), methods (3 ECTS) Example examination
• Assignment, 2.5 ECTS
  Three deliverables (see below for requirements and deadlines)
  and a presentation and active participation at the final seminar (see schedule)

Assignment

• Assignment formulation
  Structure-toxicity modeling
• Deadlines
  Deliverable 1: November 28, 2008, 3 pm
  Deliverable 2: January 12, 2009, 3 pm
  Deliverable 3: January 13, 2009, 3 pm
• Data sets
  ames_tr.zip
  ames_test.zip

Examination for DV911F (Ph.D. students only)

• Assignment, 7.5 ECTS
  An individual data analysis project is to be undertaken, which has to be presented in written and oral form. The project has to be discussed and approved in advance by the course examiner. Three deliverables are required of the same type and with the same deadlines as for the students following DV713A (see above) together with a presentation and active participation at the final seminar (see schedule for DV713A).