Using data mining methodologies to analyze patterns of student success
Date of Completion
January 2007
Keywords
Education, Administration|Education, Educational Psychology|Education, Technology of|Artificial Intelligence
Degree
Ph.D.
Abstract
The purpose of this study was to use Artificial Intelligence (AI) data mining methodologies to determine predictive variables for a cohort of students who "beat the odds" over time with respect to achievement in math or reading from Grade 4 to Grade 8. This mixed methods study was conducted within a large, urban district in the North East that is challenged by its achievement gap. ^ Quantitative data were collected using a state-of-the-art software program in AI. The program was successfully executed with classification and regression tree (CART) analyses to identify predictors of student success for this cohort of students. Qualitative data were collected in a focus group of key district administrators after presenting the CART results, exploring how useful these results were to the district, and probing what actionable steps the district could take based upon these analyses to support student academic success. ^ The district clearly understood the results of the analyses and was able to use the findings to discuss further within its own community of practice as to possible next steps it could take to intervene with students currently in elementary school, to examine why certain predictors were so strong in the models developed, and to review current assessment practices of the district with respect to outcomes of these analyses.^
Recommended Citation
Palmer, Colleen Angela, "Using data mining methodologies to analyze patterns of student success" (2007). Doctoral Dissertations. AAI3265790.
https://digitalcommons.lib.uconn.edu/dissertations/AAI3265790