[Paper Review] A visualization tool for data analysis on higher education dropout: a case study at UFES

Pedro P. Ladeira, Leandro M. de Lima+1 more|arXiv (Cornell University)|Jan 29, 2022

Evasion and Academic Success Factors4 citations

TL;DR

This paper presents an interactive data visualization dashboard built for the Federal University of Espirito Santo (UFES) to analyze higher education student dropout patterns. Using historical institutional data, the tool identifies key dropout drivers—such as socioeconomic status, course type, and geographic origin—revealing a 44.1% accumulated dropout rate (TDA) at UFES, which, while below the national average, still reflects a critical challenge for educational policy and institutional planning.

ABSTRACT

Through the analysis of cultural, socioeconomic and academic performance aspects it is possible to map the profile of the students and their motivations to drop out. This article aims to create a computational tool for data visualization that allows drawing the profile of students to support educational institutions managers in the definition of dropout avoidance policies. We present a method to treat data collected by higher education institutions over the years, analyze them to understand the dropout and provide that information to the university and the general public. Eight questions were proposed to clarify the dropout from the Federal University of Espírito Santo, Brazil. The questions were answered through the dashboard that helps to understand the causes of dropout. It is expected that this tool can be used by others educational institutions to draw student profiles contributing to possible resolution of the problem.

Motivation & Objective

• To develop a computational tool that enables educational institutions to visualize and analyze student dropout patterns using institutional data.
• To identify key socio-economic, academic, and geographic factors contributing to student dropout at UFES.
• To support university administrators and policymakers in designing targeted dropout prevention strategies through data-driven insights.
• To create a reusable, updatable dashboard accessible to both institutional managers and the public for transparency and ongoing monitoring.
• To improve data quality and visualization practices in higher education institutions by addressing gaps in data collection and representation.

Proposed method

• The study uses historical data collected by UFES from multiple academic years, focusing on socio-economic, academic performance, and geographic variables.
• A custom interactive dashboard is developed using the Bokeh library in Python to enable dynamic data visualization and exploration.
• The dashboard includes visual components such as bar charts, pie charts, and time-series plots to represent dropout trends across courses, regions, and demographic groups.
• The tool supports real-time updates as new data is collected, enabling ongoing monitoring of dropout trends.
• The system is designed with accessibility considerations in mind, though limitations in Bokeh’s functionality required workarounds using HTML and JavaScript.
• The analysis incorporates the INEP-defined Accumulated Dropout Rate (TDA) as the primary metric for evaluating dropout levels across institutions.

Experimental results

Research questions

• RQ1 What are the primary socio-economic and academic factors associated with student dropout at UFES?
• RQ2 How do dropout rates vary across different academic courses and geographic regions in the UFES student population?
• RQ3 What is the trend of accumulated dropout rates (TDA) at UFES compared to national and state averages over time?
• RQ4 How can data visualization tools improve decision-making for institutional managers in designing dropout prevention policies?
• RQ5 What improvements in data quality and visualization design are needed to enhance the accuracy and usability of dropout analysis tools?

Key findings

• UFES recorded an Accumulated Dropout Rate (TDA) of 44.1% among students who began their courses between 2010 and 2016, which is below the national average of 56.5% but still indicates a significant challenge.
• The dashboard successfully visualized dropout patterns across courses, revealing that certain programs had dropout rates as high as 67% in the dataset, highlighting course-specific vulnerabilities.
• Data quality issues, such as missing income-per-capita information and repeated entries, limited the completeness of the analysis, particularly for socio-economic profiling.
• The distance between a student’s residence and the university was identified as a potential dropout factor, though it was not fully captured due to data limitations.
• The dashboard’s functionality was constrained by the Bokeh library’s limitations, particularly in enabling interactive category selection in pie charts, suggesting a need for deeper integration with web-based visualization frameworks.
• The tool’s potential for future adaptation to include psychological and pandemic-related factors—such as mental health impacts from the COVID-19 crisis—was identified as a key area for enhancement.