Exploration of Classification Algorithms for Divorce Prediction
Danussvar Jayanthi Narendran, R. Abilash, B. S. Charulatha · Advances in intelligent systems and computing · 2020
Marital life is very important for any individual. Some marriages are successful, but nowadays many are unsuccessful. Divorce petition is filed in the family court due to various reasons. The couple faces lots of emotional and mental stress during the process. In addition, divorce gives wrong impact on the couple. The married couple ends up in divorce. Prediction is better than reaction. To avoid this situation, the comfort that will prevail among the bride and the groom is analyzed, and the success is predicted before the actual marriage. Using this prediction, the unnecessary formalities, expenses, stress can be devoid of. To analyze, the dataset is collected about the pre-marital status of the couple, which enables to predict if a marriage would be successful or otherwise before getting married. In this paper, the conclusion is drawn based on the performance of multiple classification algorithms for divorce prediction dataset. Evaluation was based on several criteria like k-fold cross-validation, mapping accuracy, sensitivity to dataset size and noise. The classification algorithms considered for the study are random forest, decision tree, XGBoost, bagging, and voting classifier.