Modeling Grade Distribution
Primary Faculty Mentor’s Name
Jebessa Mijena
Session Format
Poster
Abstract
The purpose of this research project is to see how multiple variables are associated with a student's overall performance in courses from various departments. We looked at the length, level, semester, course title and start time, in twenty-four-hour format, of the course as well as specific characteristics of the course such as the number of students and professor to see how these are associated with the distribution of grades. Our grade distributions data is collected from the Georgia College from the past seven semesters. We studied 835 sections of different courses selected from all colleges. We did not use online courses because they had no set length for the class which would make one of our variables nonexistent in that instance. In our analysis, we used both regression and tree-based models. Here are samples of a few of the results that we have found most interesting in our exploratory analysis. We found that the highest percent of W's (withdraws) was in Spring 2016 in the class PHYS 3100 which was 42.9%. Also, out of all 835 courses we studied, 18 had zero percent of A's. There were also roughly 35% of the courses we studied with less than 16 students.
Keywords
student performance, grade distribution, regression, tree-based model
Presentation Year
2017
Publication Type and Release Option
Event
Modeling Grade Distribution
The purpose of this research project is to see how multiple variables are associated with a student's overall performance in courses from various departments. We looked at the length, level, semester, course title and start time, in twenty-four-hour format, of the course as well as specific characteristics of the course such as the number of students and professor to see how these are associated with the distribution of grades. Our grade distributions data is collected from the Georgia College from the past seven semesters. We studied 835 sections of different courses selected from all colleges. We did not use online courses because they had no set length for the class which would make one of our variables nonexistent in that instance. In our analysis, we used both regression and tree-based models. Here are samples of a few of the results that we have found most interesting in our exploratory analysis. We found that the highest percent of W's (withdraws) was in Spring 2016 in the class PHYS 3100 which was 42.9%. Also, out of all 835 courses we studied, 18 had zero percent of A's. There were also roughly 35% of the courses we studied with less than 16 students.