Predictive model of school dropout in higher education through Learning Analytics with Neutrosophic Plithogenic Hypotheses

Higher education dropout rates are an international problem, as many people abandon their studies, especially in the early years. This problem affects institutions, causing significant losses. In this context, it is essential to consider methods that allow for predicting dropout rates and taking rapid and effective measures. Objective: The main objective is to analyze prediction models using Learning Analytics, integrating neutrosophic plithogenic hypotheses, to identify early the risk of dropout in higher education. This provides institutions with useful tools to implement actions that help retain more students, considering uncertainty and complex interactions between academic and non-academic factors. Methodology: The method used was quantitative, based on a selection of data from Bernardo O'Higgins University. Statistical methods with information extraction through logistic regression, factor analysis and HJ-Biplot were employed, complemented by a neutrosophic plithogenic hypothesis approach to model uncertainty in predictor variables, showing clarity regarding students who could drop out. Results: These results make it possible to identify students with problems more quickly, facilitating the implementation of specific supports from the institution, with greater precision by incorporating indeterminacy and plithogenic interactions between variables. Conclusion: The combination of statistical models with neutrosophic plithogenic hypotheses becomes a useful tool to address student dropout, allowing the development of rapid actions that contribute to educational improvement by capturing the complexity of the factors involved.