Pathways through STEM



As students go through their undergraduate careers, they take courses, interact with other students, and sometimes change their majors. Understanding what factors may act as precursors to student career choices will help us define the system that students live in and ultimately will help advising faculty understand their students. Machine learning models allow for the deep investigations necessary to understand this complex system.

Who switches to engineering from physics?

We have built a machine learning model using time stamped transcript data that predicts who will stay in physics and who will switch to an engineering program. Understanding what factors may act as precursors for student major choice will help us define the boundary conditions that students operate within ultimately leading to data driven approaches to undergraduate advising. This model demonstrates that students who switch frequently take engineering courses while registered as a physics major, and frequently do not take the third semester course in thermodynamics and modern physics.

Transfer students

A subset of MSU students transfer from different universities with credits intended towards a degree program at MSU. Reasons attributed to this behavior can be comparative cost of education (community colleges may be more affordable), perceived ease of courses at other institutions, etc. This work examines how transfer students differ from non-transfer students.

Assessing student persistence due to transformed curricula

In 2010(?) the Michigan State University Department of Chemistry transitioned their introductory sequence to the new Chemistry, Life the Universe and Everything (CLUE) reformed curriculum. This reformed curriculum was based on the design principle that students should leave the course with a solid understanding of molecular interactions (Cooper 2013). The CLUE curriculum has been shown to enhance learning (Williams 2015; Underwood 2016). While there is some evidence that increased learning happens in the CLUE classroom, there has been no evidence demonstrated that CLUE effects student persistence in the Chemistry major. The transcript data gathered as part of the MSU Pathways project stands as an excellent data set to examine this question. By comparing the pathways of students who took the traditional chemistry course to the pathways of CLUE students, we can assess the impact on student persistence CLUE has had.

References

  1. J. M. Aiken and M. D. Caballero, Methods for analyzing pathways through a physics major, 2016 PERC Proceedings [Sacramento, CA, July 20-21, 2016], edited by D. L. Jones, L. Ding, and A. Traxler, doi:10.1119/perc.2016.pr.002. [link to pdf]

Using machine learning to predict integrating computation into physics courses

We recently completed a national survey in the United States of America of faculty in physics departments to understand the state of computational instruction and the factors that underlie that instruction. We then used supervised learning to explore the factors that are most predictive of whether a faculty member decides to include computation in their physics courses. We find that personal, attitudinal, and departmental factors vary in usefulness for predicting whether faculty include computation in their courses. We will present the least and most predictive personal, attitudinal, and departmental factors.


Predicting on campus student performance from video interactions

In this work, we attempted to predict student performance on a suite of laboratory activities from students’ interactions with associated instructional videos. The students’ performance is measured by a graded presentation for each of four laboratory exercises in an introductory physics course. Each lab exercise was associated with between one and three videos of instructional content. Using video clickstream data we define summary features (number of pauses, seeks) and contextual information (fraction of time played, in-semester order). These features serve as inputs to machine learning (ML) algorithms that aim to predict student performance on the laboratory exercise presentations.

References

  1. Lin, Shih-Yin, et al. “Exploring physics students’ engagement with online instructional videos in an introductory mechanics course.” Physical Review Physics Education Research 13.2 (2017): 020138. [link to pdf]
  2. J. M. Aiken, S. Lin, S. S. Douglas, E. F. Greco, B. D. Thoms, M. D. Caballero, and M. F. Schatz, Student Use of a Single Lecture Video in a Flipped Introductory Mechanics Course, 2014 PERC Proceedings [Minneapolis, MN, July 30-31, 2014], edited by P. V. Engelhardt, A. D. Churukian, and D. L. Jones, doi:10.1119/perc.2014.pr.001. [link to pdf]