PENYUSUNAN DAN PENERAPAN METODE REGRESSION ADAPTIVE NEURO FUZZY INFERENCE SYSTEM (RANFIS) UNTUK ANALISIS DATA KURS IDR/USD

*Lamik Nabil Mu'affa  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Tarno Tarno  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Suparti Suparti  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Published: 28 May 2020.
Open Access Copyright 2020 Jurnal Gaussian
License URL: http://creativecommons.org/licenses/by-nc-sa/4.0

Citation Format:
Abstract

The exchange rate of rupiah is one of the important prices in an open economy because the exchange rate can be used as a tool to measure the economic condition of a country. The movement of the rupiah exchange rate affected the Indonesian economy, maintaining the stability of the rupiah exchange rate became an important thing to do. In an effort to maintain the stability of the rupiah exchange rate, the factors that influence it must first be identified. Several factors affect the IDR / USD exchange rate, namely the large trade price index, foreign exchange reserves, money supply and interest rates. In this study, the Regression Adaptive Neuro Fuzzy Inference System (RANFIS) method was used to analyze the effect of predictor variables on IDR / USD exchange rates. The optimal RANFIS model is strongly influenced by three things, namely the determination of input predictor variable, membership functions, and number of clusters. Determination of the optimal RANFIS model is measured based on the smallest MAPE in-sample. Based on empirical studies applied to predictor variables on IDR / USD exchange rates, it was found that the RANFIS model was optimal, namely with 3 predictor variable inputs consisting of large trade price index variables, money supply and interest rates; with the gauss membership function; 2 clusters and rules produce an MAPE in-sample of 1.93% and an MAPE out-sample of 2.68%, so the performance of the RANFIS model has a very good level of accuracy.

Keywords: Exchange rate, regression analysis, ANN, fuzzy, RANFIS

Article Metrics: