PEMILIHAN INPUT MODEL ANFIS UNTUK DATA RUNTUN WAKTU MENGGUNAKAN METODE FORWARD SELECTION DILENGKAPI GUI MATLAB (Studi Kasus: Jumlah Penumpang Kereta Api di Wilayah Jawa Non Jabodetabek)

*Tiara Sukma Valentina  -  Departemen Statistika FSM Universitas Diponegoro, Indonesia
Tarno Tarno  -  Departemen Statistika FSM Universitas Diponegoro, Indonesia
Alan Prahutama  -  Departemen Statistika FSM Universitas Diponegoro, Indonesia
Received: 13 Feb 2020; Published: 13 Feb 2020.
View
Open Access Copyright 2020 Jurnal Gaussian
License URL: http://creativecommons.org/licenses/by-nc-sa/4.0

Citation Format:
Abstract
One of the methods that is commonly used to identify a time series model and input ANFIS (Adaptive Neuro Fuzzy Inference System) model is PACF plot. The PACF plot shows the correlation between current observations and previous observations visually. Formally there are several methods that are known to effectively identify ANFIS inputs, one of which is the Forward Selection regression method. With the same concept as PACF, the process of selecting ANFIS inputs using the Forward Selection method is based on the order of the correlatiom between the predictors of the response which is indicated by the magnitude of the correlation coefficient. This study discusses the Forward Selection method in simulation data that has stationary characteristics, stationary with outliers, non stationary, non stationary with outliers and implements data on the number of train passengers in the Non Jabodetabek Java region. ANFIS modeling on data of the number of train passengers in the Non Jabodetabek Java region produces AIC of 15,5617, MAPE of 8,5093% and RMSE of 571,33691. The result of this study is equipped with a GUI which is useful as a tool to facilitate users in processing data.

Keywords : PACF Plot, Forward Selection, ANFIS, non stasionary, outlier

Keywords: PACF Plot, Forward Selection, ANFIS, non stasionary, outlier

Article Metrics: