Divorce prediction using machine learning algorithms in Ha'il region, KSA.

Area of Science:

• Social Sciences
• Psychology
• Computer Science

Background:

• Divorce is a significant societal issue.
• Artificial intelligence (AI) and predictive analytics offer novel solutions for social problems.
• The Gottman couples therapy framework provides predictors for relationship success.

Purpose of the Study:

• To evaluate the effectiveness of the Divorce Predictor Scale (DPS) in predicting divorce.
• To identify key factors contributing to divorce in the Hail region, KSA.
• To apply machine learning algorithms for divorce prediction.

Main Methods:

• Utilized the 54-item Divorce Predictor Scale (DPS) and personal information forms.
• Collected data from 148 participants (116 married, 32 divorced).
• Applied machine learning algorithms: Artificial Neural Network (ANN), Naïve Bayes (NB), and Random Forest (RF).
• Employed correlation-based feature selection to identify top predictors.

Main Results:

• The Random Forest (RF) algorithm achieved the highest accuracy rate of 91.66%.
• The study confirmed the predictive power of the DPS for divorce.
• Identified key features contributing to divorce prediction.

Conclusions:

• The Divorce Predictor Scale (DPS) is an effective tool for predicting divorce.
• Findings support the applicability of Gottman's predictors in the Hail region, KSA.
• The DPS can assist family counselors and therapists in intervention planning.