Prediksi Jumlah Perceraian Menggunakan Metode Extreme Learning Machine (ELM)

Abstract

Divorce lawsuits have considerably increased in frequency in Indonesia. According to a Statistics Indonesia estimate, there were 447,743 divorce cases in 2021, up 53.50% from the 291,677 instances that were reported in 2020. According to data from the Pekanbaru Religious Court's Public Relations, there were 1,756 divorce cases conducted in the Pekanbaru region in 2021. Extreme Learning Machine (ELM) is one of the artificial neural network technologies that can forecast. The benefit of this approach is that it has a low error rate and can train data thousands of times faster than typical feedforward algorithms. This study used the Extreme Learning Machine technique to forecast the number of divorces at Bangkinang city's religious court, where 108 divorces are expected to occur between January 2018 and December 2022. The number of neurons in the hidden layer is tested using MSE at random for hidden layer 1, 10, 50, 100, and 200 neurons. The Bangkinang religious court's divorce prediction with the lowest MSE is based on a data comparison of 80%: 20% and produces an up-and-down pattern for the number of divorces predicted for 2023: 164 in January, 66 in February, 72 in March, 74 in April, and 92 in May. If there is an increase in divorce in the upcoming month, the religious court in Kota Bangkinang can use the information that the Extreme Learning Machine can provide to come up with a solution.