Academic achievement prediction in secondary education by decision tree analysis

The aim of the present study was to develop a predictive model of academic achievement (school success or failure) by applying a decision tree analysis. A cross-sectional study was carried out to design a system for the early detection of academic failure. 219 adolescents (aged 14 to 16) participated and information on their socioeconomic status, body mass index (BMI) percentile, physical activity, leisure time spent in front of screens, enjoyment, hope, anger, anxiety, boredom, behavioral engagement, emotional engagement, cognitive engagement, self-perceived school performance and intention to go to university was collected as input variables in decision tress analysis. 6 failure and 3 success groups were found able to predict academic performance. Good accuracy was obtained in the training (80.11 %) and validation (81.40 %) datasets of the decision tree. It is possible to predict academic failure or success by assessing weight status, physical activity, anger and hope during school attendance, intention to go to university and self-perceived school performance.