Modeling the Gas price data Using Periodogram and Fourier Series Analysis

Abstract

:   This paper takes into account the application of the Periodogram and the Fourier Series Analysis. It is one of the non-parametric methods of Frequency domain analysis or spectral analysis of time series using Gas sales data in the United States of America from 1993-2014. In order to achieve these objectives، the data were obtained and then the Periodogram and the Fourier series methods were used to analyze the data. Based on the analysis، the cycle of variability within the period under study was 135 months، and a high Accuracy data model was estimated for the Fourier series which included trend، seasonal and error components. The RMSE، MASE and MAE standards were used to confirm the efficiency of the model and the model was then used to predict gas sales for six months، and we have 90% 95% confidence intervals for predictions. In addition، a time domain analysis was provided for the data series using Bok Jenkins method to obtain the appropriate ARMA model and to generate Predictions. Finally، a comparison was made between the accuracy measures of the time domain and frequency domain methods The frequency domain method competed with the time domain and the slight difference in efficiency.