An Enhanced Approach for Divorce Prediction: Implementing XGBoost Machine Learning Model with Particle Swarm Optimization on the 54 Divorce Predictors Scale Dataset

Divorce is the final termination of the marital relationship which can be caused by many things and has been increasing over the past few decades. According to this case, our paper proposes a novel study about divorce prediction using the XGBoost algorithm optimized with Particle Swarm Optimization (PSO). With the aim to enhance the performance of the divorce prediction model and provide valuable insights for effective intervention and support. Our experiment was conducted on a comprehensive Gottman dataset that consists of 54 variables known as the Divorce Prediction Scale (DPS) based on Gottman Couple’s Therapy and collected from 170 participants. With the tuned hyperparameter of XGBoost and PSO, our experiment shows the effectiveness of the divorce prediction model with the proposed method. Achieving an accuracy ranging from 98.823% to 99.411% and outperforming traditional methods also provides valuable insights into the divorce prediction area. This paper contributes to the field of divorce prediction by introducing a novel algorithm and the results show that DPS can be used to predict divorce. However, studies on this topic can be further developed by considering the amount of data and the country in which data were collected.