Pemodelan Prediksi Kinerja Kantor Cabang Bank Menggunakan Fuzzy Neural Network

Abstract

Modeling the performance prediction of bank branch unit can be a solution to predict the trend/ trend of the performance of an office so that management has the early warning system in an effort to increase the performance of an branch unit. In this study the performance of a unit is defined as the ability of the unit to reduce the ratio of NPL (Non Performing Loan) and BOPO (Operating Income compared to Operating Expenses) and raising the ratio of ROA (Return On Asset) for each month. Values decrease and increase in the ratio can be represented in a fuzzy set membership sigmoid to produce a series of time series data. Prediction model is performed using Artificial Neural Networks. Learning algorithm used was backpropagation with a sigmoid activation function. Research results obtained for the fuzzy membership model of performance and the prediction model with the performance criteria of the learning rate is optimal to apply.