DSC 433/533 - Information Analysis for Managerial Decisions

Project 1 - written

Task

In a group of two or three, analyze one of the datasets listed on the data page using one or (preferably) more of the following data mining techniques:

• Classification:
  • Logistic regression
  • Classification tree
  • Neural network (MultiLayer Feedforward)
  • k-Nearest Neighbors
• Prediction:
  • Multiple linear regression
  • Regression tree
  • Neural network (MultiLayer Feedforward)
  • k-Nearest Neighbors
• Data Reduction:
  • Hierarchical Clustering
  • K-Means Clustering

The datasets have been pre-processed to some extent, e.g., dummy indicator variables have been created for categorical features. However, feel free to change or add dummy indicator variables to suit your own particular modeling strategy. Also, note that some of the datasets contain missing data and you will need to deal with this in an appropriate way.

Use XLMiner to carry out your analysis and write up your study as a report like this. Feel free to ask me for help at any time.

Due date

Grading

Each project will be graded (with all group participants receiving the same grade) taking into account the following criteria:

• Technique: Have you used appropriate analytical techniques?
• Clarity: Have you explained clearly and concisely what you have done and why, and what your conclusions are?
• Focus: Have you met the goals of the project fully and without superfluous analyses or discussion?
• Overall presentation: Is the report presented neatly and in a professional manner?

Length: Your report must be no more than 8 pages (using a font size no smaller than 10-point). Anything beyond 8 pages will be ignored in grading. You may include a cover-page and an appendix that are not included in the page limit, but make sure that all the important material goes in the main body of the report.

Slackers: Try to resolve workload issues within your group, but if it is clear that an individual is not pulling their weight I reserve the right to reduce his/her resulting grade.