Model Hybrid SARIMA (Seasonal Autoregressive Integrated Moving Average ) - ANFIS (Adaptive Neuro Fuzzy Inference System) Pada Data Inflasi Indonesia Tahun 2003-2018
Abstract
Abstract. Inflation is is a sustained increase in the general price level of goods and services in an economy over a period of time. Inflation data measured by the inflation rate. Inflation rate needs to be stable, one way to monitor inflation is did forecasting. Basically, forecasting using the SARIMA Box-Jenkins model has given quite good results but still produces a large error value. To improve the accuracy of forecasting models, a hybrid method is carried out by combining two methods with the condition that both methods consist of linear and nonlinear components. In case of inflation data analysis, the SARIMA Box-Jenkins method is hybridized by the ANFIS method. Based on the analysis of Indonesian inflation data from 2003-2018, the SARIMA model (2,1,0) (0,0,1)12 produces 8,622045% MAPE value . Because the residual of the model is nonlinear, a SARIMA-ANFIS hybrid is carried out so that from the model obtained a MAPE value of 6.270892%. Thus, it is known that the SARIMA-ANFIS hybrid model is a better model than the SARIMA (2,1,0) (0,0,1)12 model for forecasting inflation data because it produces a smaller MAPE value.
Keywords: SARIMA, Box-Jenkis, ANFIS, Hybrid SARIMA-ANFIS, Inflation, MAPE.
Abstrak. Inflasi merupakan kenaikan harga secara umum dan terus menerus dalam suatu waktu. Data inflasi disajikan dalam bentuk laju inflasi. Laju inflasi perlu dipantau agar selalu stabil, sehingga untuk memantau laju inflasi salah satu cara yang dapat dilakukan adalah dengan melakukan peramalan. Pada dasarnya, model peramalan SARIMA Box-Jenkins telah memberikan hasil yang cukup baik namun masih menghasilkan nilai eror yang cukup besar. Dalam upaya untuk meningkatkan akurasi model peramalan, dilakukan suatu metode hybrid yaitu dengan menggabungkan dua metode dengan syarat bahwa kedua metode terdiri dari komponen linear dan nonlinear. Pada kasus analisis data inflasi, metode SARIMA Box-Jenkins di hybrid dengan metode ANFIS. Berdasarkan hasil analisis pada data inflasi indonesia tahun 2003-2018, model SARIMA (2,1,0)(0,0,1)12 menghasilkan nilai MAPE sebesar 8,622045%. Karena residu model tersebut nonlinear, maka dilakukan hybrid SARIMA-ANFIS sehingga dari model tersebut didapatkan nilai MAPE sebesar 6,270892%. Dengan demikian, diketahui bahwa model hybrid SARIMA-ANFIS merupakan model yang lebih baik daripada model SARIMA(2,1,0)(0,0,1)12 untuk peramalan data inflasi karena menghasilkan nilai MAPE lebih kecil.
Kata Kunci: SARIMA, Box-Jenkis, ANFIS, Hybrid SARIMA-ANFIS, Inflasi, MAPE.
Keywords
Full Text:
PDFReferences
Cryer, J.D & Chan, K.S. 2008. Time Series Analysis With Applications in R. Iowa: Springer.
Gujarati, D.N & Porter, D.C. 2009. Basic Econometrics. Fifth edition. New York: Mc Graw Hill.
Ispriyanti, Dwi. 2004. Pemodelan Statistika dengan Transformasi Box-Cox. Jurnal Matematika dan Komputer, Vol. 7 No. 3, 8-17.
Jang, J.S.R., Sun, C.T., Mizutani, E. 1997. Neuro-Fuzzy and Soft Computing. New Jersey: Prentice Hall.
Jang, J.S.R.1993. ANFIS: Adaptive-Network-Based Fuzzy Inference System. IEEE Trans Syst Manage Cybernet Compute, Vol. 23 No. 3, 655-685.
Kusumadewi, Sri & Hartati, Sri. 2010. Neuro Fuzzy: Integrasi Sistem Fuzzy & Jaringan Syaraf. Yogyakarta: Graha Ilmu.
Kristiana, A., Wilandari, Y., Prahutama, A. 2015. Peramalan Beban Puncak Pemakaian Listrik di Area Semarang dengan Metode Hybrid ARIMA (Autoregressive Integrated Moving Average)-ANFIS (Adaptive Neuro Fuzzy Inference System). Jurnal Gaussian, Vol. 4 No. 4, 715-723.
Makridakis, S., Wheelwright, S.C. & Mcgee, V.E. 1995. Metode dan Aplikasi Peramalan. Terjemahan oleh Untung Sus Andriyanto dan Abdul Basith (1999). Jakarta: Erlangga.
Moeeni, H., Bonakdari, H., & Ebtehaj, I. 2017. Integrated SARIMA with Neuro-Fuzzy System and Neural Networks for Monthly Inflow Prediction. Springer, 31: 2141-2156.
Nugroho, Kristiawan. 2016. Model Analisis Prediksi Menggunakan Metode Fuzzy Time Series. INFOKAM, 12(1), 46-50.
Pusat Pendidikan dan Studi Kebanksentralan. 2009. Seri Kebanksentralan: Inflasi. Jakarta: Bank Indonesia.
Qian, B. & Rasheed, k. 2005. Hurst Exponent And Financial Market Predictability. United States of Aerica: University of Georgia.
Rahman, Ahmad Zaki. ANFIS Tugas Otomasi (Online). (https://www.academia.edu/7341363/ANFIS_tugas_Otomasi/ Diakses pada Rabu, 10 April 2019)
Saputra, Arsyil Hendra. 2012. Analisis Data Runtun Waktu dengan Metode Adaptive Neuro Fuzzy Inference System (ANFIS). Jurnal Gaussian, Vol. 1 No. 1, 31-40.
Shumway, R.H & Stoffer, D.S. 2010. Time Series Analysis And Its Applications. New York: Springer.
Wei, William W. S. 2006. Time Series Analysis Univariate and Multivariate Methods. New York : Pearson.
Wiyanti, D.T & Pulungan, R. 2012. Peramalan Deret Waktu Menggunakan Model Fungsi Basis Radial (RBF) dan Autorgressive Integrated Moving Average (ARIMA). Jurnal MIPA, (Online), 35 (2):175-182, (http://journal.unnes.ac.id/sju/index.php/jm/ diakses 10 Desember 2018)
Yanti, Teti Sofia. 2010. Analisis Deret Waktu. Bandung: Pustaka Ceria.
Zhang, G. Peter. 2003. Time Series Forecasting Using A Hybrid ARIMA and Neural Network Model. Neurocomputing 50 (2003), 159–175.
Zheng, F. & Zhong, S. 2011. Time Series Forecasting Using A Hybrid RBF Neural Network And AR Model Based On Binomial Smoothing. International Scholarly and Scientific Research & Innovation 5(3), 419-423.
DOI: http://dx.doi.org/10.29313/.v0i0.18361
