Analysis of institutional data in predicting student retention utilizing knowledge discovery and statistical techniques.

摘要

With the Higher Education Student Right-to-Know Act, which requires institutions awarding federal financial aid to report graduation rates, it is increasingly important for universities to understand and study graduation rates. The selected institution, Northern Arizona University, has data assets stored in a local data warehouse. This study utilized a knowledge discovery process to build six-year graduation prediction models from these data. The knowledge discovery methodology, with its emphasis on data preparation and using multiple data mining models, uncovered attributes derived from data already being collected that could be considered valuable data assets capable of building models better than previous models based on fewer attributes.;This dissertation can be described as an ex post facto study designed to evaluate four predictive models based on available institutional data. Two of the models (logistic regression and automatic cluster detection) are common to both data mining and statistical studies. Two of the models (neural network and decision tree) are more common to data mining studies. Missing data were analyzed using proper imputation methods. Each model's predictive ability was evaluated by cross-validation in order to describe its potential usefulness in suggesting strategies to improve graduation rates.