|
Abstract:
|
From a pedagogical standpoint, one of the persistent challenges in teaching statistics is how, within an academic environment, to provide students with rich but structured `real world' experiences. Guided participation in external data challenges offers one possible solution. This year, for the MS in Statistical Practice (MSSP) program at Boston University, we integrated the JSM 2017 Data Challenge throughout a semester-long Statistics Practicum course. Teams of 2-3 students in the class formulated and executed semi-independent projects focused on analysis of the Consumer Expenditure (CE) survey data, with a focus along one of three pre-specified threads: credit risk, digital marketing, or health analytics. Tools from statistical machine learning were developed in a parallel course using just-in-time principles. Student projects were evaluated at semester end not only in the form of a course grade but also in the form of an internal competition, based on speed presentations -- with the winner of the BU MSSP internal competition to represent the program at the JSM 2017 Data Challenge.
|
