Characterization of the Dropout Student Profile Using Data Mining Techniques

Objective: One of the primary concerns in Educational Data Mining is student dropout rates. This study aims to investigate student dropout rates in higher education by identifying and analyzing the demographic and academic characteristics of university students who discontinue their studies. Theoretical Framework: Based on Educational Data Mining with clustering techniques, this study utilizes pattern recognition and data segmentation models to analyze dropout behavior within Informatics programs. Method: Data mining techniques were applied to a dataset that contained demographic and academic records. Three clustering algorithms, K-means, DBSCAN, and Agglomerative Hierarchical Clustering, were employed, and their performance was evaluated. Results and Discussion: The K-means algorithm produced three distinct clusters with a silhouette score of 0.575, indicating well-defined groups. These clusters revealed significant patterns, such as a predominance of male, single students from Cusco enrolled under the 2013 curriculum. DBSCAN identified four clusters (score: 0.120), while Agglomerative Hierarchical Clustering produced three clusters (score: 0.564), offering a balance between granularity and clarity. The findings highlight the effectiveness of K-means in profiling dropout students and offer insights into their academic trajectories. Research Implications: The findings suggest that tailored interventions addressing the specific needs of identified student clusters may reduce dropout rates, relying on informed policy and practice in higher education. Originality/Value: This study contributes to the literature by introducing an innovative comparative analysis of clustering methods for dropout profiling, offering practical implications for educational data analysis and intervention strategies.