MODEL HYBRID SINGULAR SPECTRUM ANALYSIS DAN NEURAL NETWORK UNTUK PERAMALAN KENAIKAN NILAI INFLASI DI INDONESIA
No Thumbnail Available
Date
2023-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Elfitra
Abstract
Current economic developments cause increasing inflation rates in a country. One of the
statistical methods used to determine the increase in inflation values is forecasting using
a non-parametric time series model. This research was carried out using Singular
Spectrum Analysis and Neural Network as a non-parametric forecasting method with
monthly data on inflation values in Indonesia from January 2003 - December 2022. This
analysis was carried out by forming a square matrix from the research data so that
eigenvalues and eigenvectors were obtained in each matrix. as many as 50. In the
calculations, the forecast results obtained for the next 5 month period show insignificant
increases and decreases. Based on the accuracy results, an error was obtained using
MAPE with forecasting results for the inflation value of 9%, which can be said to be in
the very good category.
Description
Keywords
Inflation, time series, Singular Spectrum Analysis and Neural Network
Citation
Perpustakaan