Machine Learning can predict Satisfaction in Relationship

The most reliable prediction of a relationship’s success is the partners’ belief that the other person fully commits. That is what a Western University-led international research team has found.

Other critical factors in a successful relationship include being appreciated, feeling close to, and sexually satisfied with your partner. This is what the study says. The study was made by the first-ever systematic attempt at using machine-learning algorithms to predict relationship satisfaction of people.

Samantha Joel is a Western Psychology professor. She said that satisfaction with romantic relationships has important implications for work productivity, health, and wellbeing. Nevertheless, research on predictors of relationship quality experiences limitations in scale and scope. They carried it out separately in individual laboratories.

Paul Eastwick from the University of California, Davis, Joel, and 84 other scholars from around the world conducted an extensive machine-learning study. It delved into 43 distinct self-reported datasets on romantic couples for more than 11,000 couples.

Relationship Success

Proceedings of the National Academy of Sciences published this comprehensive study. Thus, the study provides provisional answers to the perennial question: ‘what predicts how happy I will be with my relationship partner?’ Joel used an application of AI (artificial intelligence), machine learning, to comb through vast combinations of predictors. It was far more than a single researcher could hope to analyze in a lifetime. This was to find the most reliable and robust predictions of relationship satisfaction.

Relationship-specific predictors like ‘sexual satisfaction,’ ‘perceived partner commitment,’ and ‘appreciation’ account for nearly half of variance in relationship quality, the study concluded.

Individual characteristics describe a partner rather than a relationship. So, it explains 21 percent of the variance in relationship quality. The top five personal attributes with the most reliable predictive power for relationships are ‘anxious attachment, ‘avoidant attachment,’ ‘depression,’ ‘negative affect,’ and ‘satisfaction with life.’

Joel said that relationship-specific variables were about two to three times as predictive as individual differences.