اميرمحمد پاس لاري

Education, as one of the fundamental pillars of human development, plays a decisive role in individual an‎d social growth. Studentsʹ academic performance is directly related to family an‎d social factors. This study aimed to model students’ academic underachievement based on family characteristics through a case study in Alborz province. The data used in this research were collected from primary an‎d secondary schools in Alborz province an‎d included variables such as parentsʹ education level, household income, parentsʹ occupation, an‎d family structure. In this study, four machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM) with RBF kernel, Ran‎dom Forest, an‎d XGBoost, were employed to design an‎d implement models for predicting academic underachievement. Additionally, GridSearch was applied for hyperparameter optimization to improve the performance of the models. To investigate the relationship between family characteristics an‎d academic underachievement, Spearman correlation was used as a non-linear statistical method. The SMOTE technique was also utilized to han‎dle data imbalance an‎d balance the classes. The findings indicated that the mother’s education level, family income, an‎d father’s education level were the most influential factors in predicting students’ academic underachievement. Among the different algorithms, the Ran‎dom Forest model provided more accurate predictions an‎d better capability in simulating complex relationships, successfully identifying students at risk of academic underachievement. Furthermore, SHAP analysis was applied to identify feature importance, revealing that mother’s education level an‎d family income had the greatest impact on predicting academic underachievement. The results of this research can serve as an effective tool for predicting the risk of academic underachievement an‎d for designing targeted educational policies, particularly to support students at risk of poor academic performance.

ويژگي‌هاي خانوادگي , يادگيري ماشين

Family Characteristic , Machine learning

Amir Mohammad Paslari

Javad Vahidi