Salt Lake City, UT
June 24 – 27, 2018
The American Society for Engineering Education annual conference is committed to fostering the exchange of ideas, enhancing teaching methods and curriculum, and providing prime networking opportunities for engineering and technology education stakeholders.
Tuesday, June 26
Visualizing Design Team Analytics for Representing and Understanding Design Teams’ Process
Corey T. Schimpf, Rob Sleezer (Minnesota State University, Mankato), Charles Xie
1:30 – 3:00 PM, Room 255 E, Convention Center – Salt Palace
Engineers typically approach design in teams, particularly when dealing with complex problems that may need to be decomposed into several parts or subsystems to be designed individually and integrated. Team design projects during students’ college years can serve as critical experiences to prepare for professional work on design teams. However, the volume of actions across team members and iterative nature of engineering design makes tracking, representing and learning from design teams’ actions difficult and time-consuming. This work proposes developing design team analytics as a tool for representing and understanding how students collectively navigate and address complex designs, by leveraging a computer-aided design (CAD) platform with action-logging functionality.
A class of 28 juniors and seniors in a project-based engineering program at a small Midwestern University worked in teams of four-to-five to design a distributed system of solar arrays for their local community while balancing energy need and budgetary constraints. Students were given a suite of 8 solarizable sites including flattop, pitched-roof buildings, and parking lot locations. Students then used these sites to design, evaluate, and select a subset for their final design. Energy3D, a CAD platform for constructing buildings and solar arrays that features many analytical tools, served as the primary design platform. Importantly, Energy3D logs users’ actions such as adding a solar panel or running an annual solar yield. The data from these logs was examined in terms of individual and collective contributions resulting in visualizations of the teams’ design processes across several metrics including: construction, optimization, and numerical analysis.
Preliminary results for this work-in-progress indicate that students mostly designed sequentially across solarizable sites, with little concurrent activity. Optimization patterns vary between teams and show some relation to teams’ final design(s) performance.
Wednesday, June 27
A Markov Chain Method for Modeling Student Behaviors
Corey T. Schimpf, Molly H Goldstein (Purdue University-Main Campus, West Lafayette (College of Engineering)), Robin Adams (Purdue University-Main Campus, West Lafayette (College of Engineering)), Jie Chao, Senay Purzer (Purdue University-Main Campus, West Lafayette (College of Engineering)), and Charles Xie
11:30 AM – 1:00 PM, Room 150 E, Convention Center – Salt Palace
Students from a middle school (N=152) and from a high school (N=33) completed the same energy-efficient home design challenges in a simulated environment for engineering design (SEED) supported by rich design tool with construction and analysis capabilities, Energy3D. As students design in Energy3D, a log of all of their design actions are collected. In this work-in-progress a subsample of the five most engaged students from both the middle and high school samples are analyzed to identify similarities and differences in their design sequences through Markov chain models. Sequence learning is important to many fields of study, particularly fields that have a large practice component such as engineering and design. Design sequences represent micro-strategies for developing a design. By aggregating these sequences into a model we aim to characterize and compare their design process. Markov chains aid in modeling these sequences by developing a matrix of transition probabilities between actions. Preliminary results suggest we can identify similarities and differences between the groups and that their design sequences reflect important considerations of the design problem. We conclude that Markov chains hold promise for modeling student practices.