Modeling the Nexus between Parenting Style and Anti Social Behavior using Ensemble Learning Approach

Abstract

Contemporary society is grappling with issues of anti-social behavior in children and adolescents, one of which is influenced by parenting styles. This research employs machine learning technology, particularly ensemble learning, to model the relationship between parenting styles and anti-social behavior. The research data is derived from previous studies encompassing parenting style parameters and anti-social behavior. This data is preprocessed and feature-engineered, then used in modeling through the Random Forest (RF) and Adaptive Boost (AdaBoost) methods. Modeling is conducted in two phases: vanilla modeling and hyperparameter tuning. The results of the tuned models indicate that RF performs better (accuracy=91%) than AdaBoost (accuracy=72%). In conclusion, RF, as a bagging ensemble learning technique, effectively models the relationship between parenting styles and anti-social behavior. Future studies are recommended to gather more training data and develop an early detection system for use by child psychologists in the field.