Description

The term project is a group project intended to allow students to explore some aspect of data mining deeply. The nature of the project is determined by each group but must meet the following requirements:

• The project must explore some aspect of data mining deeply.
• Each group member must contribute significantly to the implementation of the project (e.g., it's not acceptable for the only role of a group member to be the web page maintainer).
• The project must be challenging; that is, it must not be immediately evident to the group members how the project is to be implemented.

Because the project proposal is submitted during the first half of the semester, you will have to skim ahead in the textbook to identify interesting topics or approaches to data mining. Furthermore, you will have to read about your topic more deeply once the basic idea for your project is chosen.

Each group may (and is expected to) discuss topic ideas with the instructor and ask for direction during the proposal formulation stage. We will offer guidance and clarify any questions as each group prepares its proposal.

Deadlines

Point Distribution

Project Web Site

• A brief overview of your project.
• Your project proposal.
• An up-to-date progress report for the project.
• Links to software being used.
• Instructions for obtaining your project.
• Instructions for building your project.
• Instructions for executing your project.
• Your progress report.
• Your final project report.

You should take the web site creation and maintenance into consideration when determining the amount of time the project will require.

When your project web site is available, contact the instructor with the URL for your project site.

Project Ideas

• A complete case study of a particular application domain for data mining. This would involve feasibilty analysis, data cleaning, data mining, and reporting/visualizing your results. Light programming.
• Add data mining features to a publicly available (open source) data base system like postgres, mSQL, or MySQL. This would involve selecting a query language syntax for the data mining feature(s) and performance analysis of your implementation. Medium to heavy programming.
• Research and develop improved implementations of data mining algorithms, preferably implementing them as part of the Weka environment. This will also involve performance analysis. Medium to heavy programming.
• Research and develop tightly integrated data mining algorithms with commercially available database packages. Medium programming.
• Other projects that look to improve or enhance data mining efficiency, scalability, or understandability.
• Look into "closing the loop". Knowledge acquisition, including data mining, is an interative process with no precise end. A project in this area would investigate ways to further automate this iterative process, feeding the output from one data mining task into the input of another.

Proposal