Reference no: EM131816844
Classification is a useful human skill. Imagine walking into your favorite clothing store and seeing all of the clothes piled together on a center table. T-shirts and pants and socks intermingle, with the sizes mixed up.
Retail stores organized like this would not survive, nor would distributors or manufacturers who managed their inventories this way. Sorting and classifying are necessary, important, and essential activities. But those activities can also be dangerous. Serious ethical issues arise when we classify people. What makes someone a good or bad "prospect"?
If we're talking about classifying customers in order to prioritize our sales calls, then the ethical issue may not be too serious. What about classifying applicants for college? As long as there are more applicants than positions, some sort of classification and selection process must be done. But what kind? Suppose a university collects data on the demographics and the performance of all of its students.
The admissions committee then processes these data using a decision tree data mining program. Assume the analysis is conducted properly and the tool uses statistically valid measures to obtain statistically valid results. Thus, the following resulting tree accurately represents and explains variances found in the data; no human judgment (or prejudice) was involved
Discussion Questions
1. Explain what conditions in the data could have caused this particular structure to emerge. For example, what conditions may have existed for self-funding students under the age of 23 to be classified as low risk? Explain how you think the three other branches in this tree may have come about.
2. Consider this tree from the standpoint of:
a. A 23-year-old woman whose job experience is 3 years as a successful Wall Street financial analyst.
b. A 28-year-old gay male with 4 years' job experience who has no children and pays his own college education.
c. The university fund-raising committee that wants to raise money from parent donations.
d. A student who was seriously ill while attending a top-notch high school but managed to graduate with a GPA of 2.9 by working independently on her classes from her hospital room.
3. Suppose you work in admissions and your university's public relations department asks you to meet with the local press for an article they are preparing regarding your admittance policy. How do you prepare for the press meeting?
4. Would your answer to question 3 change if you work at a private rather than public institution? Would it change if you work at a small liberal arts college rather than a large engineering-oriented university?
5. What conclusions do you make regarding the use of decision trees for categorizing student applicants?
6. What conclusions do you make regarding the use of decision trees for categorizing prospects in general"?