dimacs logo

General InformationProject DescriptionProject LogLinks

General Information

Brandon Blakeley
CoRE 446
The University of Texas at Austin
Faculty Advisor:
Endre Boros, Operations Research
Minimum Cost Optimal Decision Tree

Project Description

In many situations costly tests lead to a final diagnosis. Based on past data, it is a hard problem to find the best diagnosis based on the attributes. Even harder to find the one which provides the best diagnosis at the smallest expected cost. In this project, we evaluate and extend a dynamic programming algorithm solving this problem.

Presentation 1 (ppt)

Presentation 2 (ppt)

Project Log

Week 1: Met mentor, realized he was going to be out of town for 5 weeks, sought new mentor

Week 2: Found new mentor and selected project

Week 3: Read background information.

Week 4: Familiarized myself with code, manually tested with toy data sets

Week 5: Developed automated testing framework

Week 6: Continued developing automated testing framework

Week 7: Composed presentation

Week 8: End of program

Links and Resources

Rutgers DIMACS

McKinney et al. "Machine Learning for Detecting Gene-Gene Interactions"

Guyon, Weston, and Barnhill. "Gene Selection for Cancer Classification using Support Vector Machines"

Gunn, Steve. "Support Vector Machines for Classification and Regression"

Top of Page