ACCT 6500 Acct Information Systems II (UACPA)

ACCT 6500: Course Embedded Assessment

UACPA Fall 2006

Percentage of Students Meeting Learning Objective's Performance Expectation.

Learning Goals	Student Performance
1. Each student will be able to correctly interpret the lift chart of a data mining decision-tree model.	96%
2. Each student will be able to correctly apply the results of the data mining model to a specific business decision.	72%
3. Each student will be able to demonstrate abstract data modeling skills by constructing an associative schema from an application description.	72%
4. Each student will be able to demonstrate the application of the associative data structure for transactions based on the schema in No. 3 above.	92%

Analysis and Recommendations: How might the above learning activities be improved to raise student performance levels? How might you change outcome objectives and/or assessment methods based on the above results? Other observations and/or recommendations?

Learning Outcome Objectives 1 and 2. Students generally had little difficulty interpreting the lift-chart curve. However, several students stumbled when asked to quantify the potential savings from employing a hypothetical data mining model. Most of these errors seemed to occur because students confused the target respondents with the population. This is surprising because the lift charts in the data mining software (KnowledgeSeeker) clearly mark the x-axis as the "target respondents." The axis is labeled "% responded" on the exam. The students who missed this portion seemed to equate the x-axis (target) with the y-axis (population). It would appear that more emphasis should be placed on contrasting the meaning of these two axes. In addition, the exam should refer to the x-axis using the same labeling terminology as the data mining software to eliminate another source of confusion.

Learning Outcome Objectives 3 and 4. The students who generated errors on the associative schema seem not to grasp the associative entity re-use concept. For example, when modeling the "date" scalar, these students created multiple date entities (example: hire-date, birth-date, termination-date) instead of drawing different associations to a single date entity. Others seemed not to understand that a person can have multiple roles in the system (example: customer, vendor, employee) yet only be represented once in the system, unlike the relational model, where the same person might be entered in several master tables (example: vendors, employees, customers). It would appear that having the students work through a couple of comparative examples (i.e., render the same data in both the associative and relational data models) and contrast the way each model represents the data.