Predicting and Analysing Adult Learners Dropout Indicators From Literacy Programmes Using Data Mining Approaches

Abstract Data mining provides effective techniques for discovering useful knowledge and patterns from students’ data. The discovered patterns can be used to understand many problems in the educational field. One of the challenges facing academic institutions is dropout student issues. It is important to recognize which students are at risk of dropping out of school and what the underlying factors are. This study aims to predict adult learners’ dropout and analyses the causative factors that lead to the dropout of such learners using data mining techniques. The dataset used for this research was collected from 97 adult learners who registered with the Adult Literacy Program offered by the Kwara State Agency for Mass Education in four selected centers in Offa for the academic years 2019, 2021, and 2022. Three decision tree algorithms, C4.5, ID3, and CART classification methods, applied to the student’s socioeconomic and demographic data using the WEKA Data Mining tool to produce the best prediction model for learners dropping out of the ALP. Researchers conduct experiments to identify the best model among the techniques used, and then calculate the accuracy of the models. The findings from the study suggest that ID3 was the best algorithm with an accuracy of 80%, compared to C4.5 with an accuracy of 79.5% and CART with 78.5%. The experiment result also shows that some of the top indicators for learners’ dropout were age, relevance of the program, and number of working hours. This study suggested that future research should integrate study-related information with socioeconomic and demographic indicators for predicting learners' dropout rates.