4. Consider the following dataset, which includes three attributes and a classification for admission decisions into an MBA program:

a. Using the data given here as examples, develop your own manual expert rules for decision making.

b. Build a decision tree using SPRINT (Gini index). You can build it by using manual calculations or use a spreadsheet to perform the basic calculations.

c. Build another decision tree, using the entropy and information gain (ID3) approach. You can use a spreadsheet calculator for this exercise.

d. Although the dataset here is extremely small, try to build a little neural network for it.

e. Use automated decision tree software (e.g., See5;
download a trial version from rule-quest.com) to build a tree for these data.

f. Report the predictions on the last three observations from each of the five classification approaches.
g. Comment on the similarity and differences of the approaches. What did you learn from this exercise?

Transcribed Image Text:

Quantitative GMAT GPA GMAT Decision 650 2.75 35 NO 580 3.50 70 NO 600 3.50 75 YES 450 2.95 80 NO 700 3.25 90 YES 590 3.50 400 3.85 640 3.50 540 3.00 8 8 80 YES 45 NO 75 YES 60 ? 690 2.85 80 ? 490 4.00 65 ?