Predict Students' Dropout and Academic Success

Team Members: 

Ethan Yitzhaki 

Sophie Rhodes 

Torin Brubaker

Townsend Powell 

Summery: 

Colleges like to determine if their students will be successful or not and sometimes the easiest way to do it is with simplified data. This dataset's goal was to help examine the dropout and success rate of students in higher education. It does this by looking at 36 different attributes across 4,424 students who were a part of this study. To study this data set we used the Lindsay Poirier methodology of a denotative, connotative, and deconstructive readings in order to answer the questions, What is this data useful for in regards to higher education? What are its limitations in looking at students? 

We first conducted a general quantitative analysis to determine the risk factors and predictors of success represented by the data. Then we broke the data down into categories, including academic metrics, socioeconomics, and demographics information. Lastly, we completed the Poirier three-step analysis to help us fully understand the data and synthesize it with the potential predictive use of the data.

For the denotative reading the dataset is composed of 35 variables divided into graduate, enrolled, and dropout students. The data is then analyzed into which factors pertain the biggest influence on success or failure in high education. For the connotative reading, we evaluated the cultural, political, social, and institutional implications of the data, particularly through the context of the data collection company being based in Portugal and the underlying meaning of determining student success through the labels of either graduated, enrolled, or dropped out. Through the deconstructive reading, we determined that, while the data set includes key data for educational institutions, it is not a holistic range of data and does not include significant metrics like student extracurricular involvement, whether students had accessed mental or physical health resources, or students’ employment status. Additional data can provide a stronger understanding of students’ risk factors that expands beyond markers of already-occurring academic struggles, and this can be used to direct institutional resources towards those students. In conclusion, we found that the data can be useful for institutions but would best support the goal of encouraging student success by integrating any predictive algorithm developed using the data with a holistic and preventative approach that prioritizes supporting students’ needs.