GARUT -In the development of statistics, regression analysis is a very important analysis in all fields. Especially the field whose goal is to see the relationship between variables. In the world of statistical regression analysis continues to be developed in considering the acquisition of better results in modeling. So that today many regression analysis approaches have been introduced. One of them is used to model inflation in the food sector in Indonesia with time series data. It is possible that the modeling carried out by a study can be influenced by various factors. One of the factors that influence the modeling results is the presence of data patterns in the data used.
From my point of view, modeling inflation in the food sector in Indonesia can be overcome by multiple linear regression analysis. However, the use of this analysis is considered inaccurate because it only looks at the relationship between variables without considering the use of time series data used in the study. The use of time series data can increase the likelihood of data pattern constraints. So, in dealing with this, it is necessary to use the right analysis. Errors in using the analysis can have an impact on unsatisfactory modeling results. For this reason, it is necessary to identify an analysis that can overcome data patterns that occur due to the use of time series data used in research.
Based on the article Fourier regression modeling for case study time series data: inflation modeling in the food sector in Indonesia by Tiani Wahyu Utami, 2018. Modeling of food sector inflation in Indonesia was studied using regression with the Fourier series approach and multiple linear regression analysis. The Fourier series is a nonparametric regression approach that is able to overcome data that has a trigonometric spread due to the influence of patterns on the data used. Meanwhile, multiple linear regression analysis is an analysis that looks at the relationship between variables. This article compares the Fourier series analysis and multiple linear regression in modeling inflation in the food sector in Indonesia. The data used is in the form of monthly data from January 2007 to August 2017. After statistical tests were carried out, it was found that the Fourier series analysis had better modeling results than multiple linear regression analysis. It is proven that the multiple linear regression model produces R-Square 90%, while for the Fourier series it produces 99% R-Square.