Predictive Model of Student Dropout Based on Logistic Regression

Abstract

Abstract Student desertion is a phenomenon that has spread significantly in many higher education institutions in Ecuador. The objective of the research was to develop a predictive model of student dropout based on multiple binary logistic regression, with the purpose of detecting possible dropouts. The methodology used consists of three phases: Phase 1: Analysis of variables; Phase 2: Formulation of the mathematical model; and Phase 3: Evaluation. For the estimation of the coefficients of the model, the SPSS tool was obtained. After the creation of the predictive model, it was concluded that the most significant variables that contribute to the diagnosis of dropout are marital status, age, gender, Note2s, and Note1s. It is also evident that students have a higher risk of dropping out if they are married and lower risk if they are single or divorced. Finally it was concluded that gender is a factor that directly influences dropout; male students are more likely to drop out than females.

Keywords

• logistic regression, predictive model, desertion.