Contemplative Course Design & Implementation: A Case of Senior Level Data Mining Course
As much as a college professor relies on their past teaching experience and training, one should also reflect on their students’ years when teachers themselves were the consumers of what they are delivering now in classrooms. By reflecting on those years, teachers may ask themselves such questions as: why did I like some courses over others, some teachers more than others? Why didn’t I like some courses at all? And what do I recall from each of those many courses that I had to take during college years? Contemplating these questions and answers to them can become a guide in developing current courses and teaching methods. Teachers may find not all that they have learned in classrooms and through courses can be recalled now after many years but only a few bits and pieces from each class. Typically, if one is asked about a specific course, they can recall what it was about and point out a few things unless they have frequently used the learned material in their professional life after graduation. People rarely remember the details of methods or tools used in driving solutions. Still, they can generally recall the general process of identifying a problem and arriving at a solution to that problem.
Reflecting on my student years, I have figured what has made a course enjoyable or distasteful to me and what has had lasting effects on me. Drawing on the results of such reflections and contemplations, I realized that generally, I could recall the significant points and principals of many courses I took during my undergraduate education. I also could remember more details from some courses where I struggled to master the material mostly on my own (self-learn) instead of only relying on the lectures and textbooks. Consequently, it became of paramount importance to me to find clear answers to the following questions before designing a course: What are the major points and processes that need to stay with students long after their graduation? What elements are suited for self-learning that will give students the most value? Once the major points and logical processes are identified for a course, then they can be transformed into a course theme and that lends itself to driving the course learning objectives and course material (lectures, assignments, quizzes, exams, discussions, etc.) design. Finally, aspects of the course that are better fitted for self-learning (e.g. learning tools as opposed to learning logic) will be identified and implemented as such. To illustrate further, I will describe the design and implementation of an elective data analytics course at a major business school located in the southwest region of the United States.
Data analytics and data mining are rooted in different disciplines (e.g., computer science, statistics, business, etc.), and hence they can be taught from different perspectives and focal points. Being a college of business course, therefore, the course should be taught from a business application perspective as opposed to computer science or statistics. Naturally, I adopted the Cross-Industry Standard Process for Data Mining (CRISP-DM) to become the theme of the course. CRISP-DM is a logical process that aims at encompassing varied approaches towards data mining projects and is developed by a consortium of industry leaders. Hence, this process model was adopted to become the core of this course, the one knowledge that I hoped to stay with my students long after they graduate. The CRISP-DM is a life-cycle model geared towards a successful identification of problems, opportunities, and goals of data mining projects and eventually delivering value through actionable business solutions. It consists of six phases that start with business understanding and data understanding. Problems, opportunities, and goals of a project are identified and formulated in these two phases as one informs the other and vice versa. Then the process moves to data preparation and modeling phases. These two phases are also interdependent as the choice of modeling technique may introduce some data preparation constraints. These phases also depend on the previous steps since the choice of modeling technique itself is influenced by the business goal, data availability, and data specifications. Next is the evaluation phase, where the analyst needs to determine if the results of the modeling effort are good enough to proceed forward to the deployment phase, where actionable business solutions are developed and deployed depending on the project goal.
Accordingly, the course objectives are developed using the CRISP-DM process, and students are required to follow the process and report on each phase in their assignments and the course project. This course has eleven assignments where the CRISP-DM is the central theme of the eight of them. About twenty percent of assignments grade is reserved for students’ adherence to the process and the mastery of it. Students also get tested on their knowledge and mastery of the process during the midterm and final exams. As the main theme of the course, the CRISP-DM is repeated and reinforced over and over during the semester through lectures, assignments, exams, and course project with the hope of getting it ingrained on the minds of students while they learn about different data manipulation and modeling techniques (e.g., outlier detection, data scrubbing, association analysis, a multitude of regression and classification analyses, etc.) and tools.
To determine the aspect of the course that is better fitted for self-learning, I look into industry trends to identify tools and techniques that their knowledge and mastery will increase the chance of students finding a job. The course project, which is a group activity, is typically the best fit for the self-learning aspect of the course. Not only students learn a tool on their own, but they also will learn from each other in the process, and some of them may find joy in teaching others what they have learned. For this aspect of the course, I provide references to appropriate and useful learning contents for the tool and provide a structure for the project and guidance on business objectives. If there are free online certification/training programs for the tool, students are often encouraged to submit an individual certificate of training completion along with their final project report for some extra credit. In addition to boosting students’ self-efficacy in learning new software tools, this practice helps them to stay relevant in a highly competitive and dynamic data analytics market where new tools and methods are surfacing up at a rate never seen before.
This approach to course design and implementation seems to be working very well, as evidenced by the student evaluations of me and the course and through students' comments over more than seven years of teaching this elective course and continuous improvements of it. I would be happy to share my experience in developing and deploying a central theme for courses as well as this course syllabi and course materials with interested readers upon request.
Notes:
An earlier version of this post has appeared on the Coles College of Business faculty blogs webpage.
