Publication

Abstract : Divorce rates are increasing around the world. During the procedure of divorce, the couple is under a lot of emotional and mental strain. Moreover, divorce also has social and economic ramifications. Considering these points, it is important to examine and analyze the divorce phenomenon. To reduce and prevent the occurrence of future divorce cases, it is very crucial to know what are the factors which lead a couple to divorce. In our proposed work, we have primarily used four classification algorithms namely Naive Bayes, Random Forest, KNN, and GLM then further applied the data mining technique of MCA on KNN and GLM to improve the performance of divorce prediction. The models were evaluated based on Leave-One-Out-Cross-Validation (LOOCV). We have been able to achieve the least test error of 0.58% with GLM + MCA and KNN + MCA.