Discussion: View Thread

Learning Action Awards - Redesigning Business Analytics

  • 1.  Learning Action Awards - Redesigning Business Analytics

    Posted 07-30-2020 10:18:00 AM

    BADM 211 Business Analytics II is a core course offered in the Gies College of Business at the University of Illinois and it is designed for business undergraduates as their primer into machine learning and data-driven business analysis. It should be noted that, despite efforts (e.g., Mashayekhy & Yetgin, 2020), there is no model curriculum for business analytics coursework (Stephens & McGowan, 2018), and because of the topics involved therein (mathematics & statistics, computer science, and business analysis), a variety of approaches may be employed: math & statistics, computer science, or business-centric.

    While using a code-first teaching approach is discouraged from the perspective of enhancing diversity in the classroom or program (National Academies, 2018), based on our experiences from last semester regarding the difficulty students had learning after the course was moved online due to a national emergency, for Fall 2020 the code will be used as the anchoring proxy for the data mining concepts. This teaching model should work well because the prerequisites for BADM 211 are one semester of Python programming (CS 105) and one semester of college algebra (MATH 115) – one semester of Python will better prepare a student for the code-first teaching model than one semester of college algebra would prepare a student for a math-first approach.

    Cognitive load theory was consulted during this redesign because of the nature of the course and its elements of mathematics & statistics, computer science, and business analysis, each with a different cognitive load. Some researchers into cognitive load theory have developed a heuristic that states when the topic contains both concepts and technology, teaching technology skills followed by conceptual underpinnings will force a higher level of interactivity and result in better learning outcomes for students with low-technology skills (Kaluga, et al., 2003; van Merriënboer, et al., 2003). While our course redesign supports this tactic, measuring cognitive loads and devising prescriptive interventions for individual students is still out of reach until we have a data mining process to achieve that result.

    With a code-first method, all concepts will be anchored by the Jupyter notebooks (coding environment), and the new design is now suited for live, online, or hybrid coursework. This new tactic will require a concept to be explained and then shown how it is implemented with a detailed description of every code element, and then the students will have to complete a related code segment with deliberate errors or omissions and get it to run properly. This will force a back-and-forth flow in the live or asynchronous session, with no more than 5-15 minutes spent in exposition before the students are asked to engage. Another upside to this code-first direction is that it better prepares business students with coding skills that employers are requesting (Radovilsky, et al., 2020). There is a drawback to a code-first approach and that is the fact that code changes. What took 100 lines of code several years ago only required 10 lines of code last year, and by next year, perhaps a single line of code. But, we are using the code to anchor basic to intermediate data analytic concepts and some math, and those concepts are fairly static.

    These lecture snippets, which cover topics such as linear regression, regularization, dimension reduction, and resampling among many more, will be pre-recorded as asynchronous videos with the students then asked to pause the video, complete the micro-assignment in the Jupyter notebook, and then restart the video for more information and another request to engage. The homework in this course will still be fully asynchronous with auto-graded assignments that put a higher weight on data analysis instead of coding because lectures are now coding sessions. Due to the hybrid/online environment, assessments will be of the open-book, open notes variety but with a tight time limit (60 questions, 60 minutes). We tested this assessment structure last semester, and it worked well in terms of student difficulty balanced against the potential for academic dishonesty.

    Once all the redesign components are complete, we believe that this course will readily adapt to any learning environment, whether live session, hybrid, or fully online.

     

    References

    Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). Expertise reversal effect. Educational Psychologist, 38, 23–31.

    Mashayekhy, M., Shen, J., & Yetgin, E. (2020). Toward Designing a Business Analytics Model Curriculum for Undergraduate Business Students.

    National Academies of Sciences, Engineering, and Medicine. (2018). Envisioning the data science discipline: the undergraduate perspective: interim report. National Academies Press.

    Radovilsky, Z., Hegde, V., & Damle, L. H. (2020). Data Mining in Business Education: Exploratory Analysis of Course Data and Job Market Requirements. Journal of Supply Chain and Operations Management18(1), 119.

    Stephens, P., & McGowan, M. (2018). ISSUES IN THE DEVELOPMENT OF AN UNDERGRADUATE BUSINESS ANALYTICS MAJOR. Issues in Information Systems19(2).

    van Merriënboer, J. J. G., Kirschner, P., & Kester, L. (2003). Taking the load off a learner's mind: Instructional design for complex learning. Educational Psychologist, 38, 5–13.

     



    ------------------------------
    David Guggenheim
    Teaching Assistant Professor
    University Of Illinois - Urbana-Champaign
    Champaign, Illinois
    872-2050574
    ------------------------------