2012 ISECON Proceedings
New Orleans LA
Exposing Students to Big Data: Hands-on Market Basket Analysis with Access
Market Basket/Affinity Analysis can be used to help retailers identify products purchased together where the association might not be obvious, i.e., hidden associations. These discoveries can be used to better understand consumer behavior to drive the sale of associated items. In our Market Basket Analysis (MBA) activity, students engage in data mining where marketing knowledge is discovered by transforming large amounts of transaction data into association rules. While such activities might be common in a Data Mining course, our MBA activity is appropriate for an introductory business database, computer applications, or management information systems course. Our MBA activity can be performed using only the query design tool in Microsoft Access where knowledge of SQL is not a prerequisite. We use a publicly available retail database where the Apriori Algorithm allows for the computation of itemset quadruples that would otherwise be intractable. Thus, students see the importance of pruning in efficiently handling big data. Using a worksheet activity, we briefly review MBA concepts including itemsets, association rules, support, confidence, and quality. Then, using Access (hands-on), we guide participants through an MBA activity using a database with over 900,000 rows. We also present an alternative, SQL-based approach to creating itemsets and association rules. We provide enough detail so instructors can adapt our activity to different settings. In a lecture, instructors can use the worksheet to engage students and then demonstrate the MBA activity. In a lab, novice students can conduct the activity using step-by-step instructions, or advanced students can independently create many of the queries using less detailed instructions, in either the SQL or non-SQL scenario. The worksheet, instructions, and database will be available to participants, who should bring a laptop running Access 2007/2010, or be able to share such with at most one other attendee.
This workshop is appropriate for faculty interested in using a realistic, hands-on Market Basket Analysis activity to teach data mining principles and both introductory and advanced query design principles using Access’s query design and/or SQL. The activity can be used in a lab, classroom (with laptops), or as homework.
Recommended Citation: Breimer, E., Vandenberg, S., Yoder, R., (2012). The Proceedings of the Information Systems Education Conference, v.29 n.2061, New Orleans, LA.