The application of Seasonal Trend Decomposition Using Loess for Export Forecasting by Economic Commodity Group in North Sumatra

ABSTRACT

1. INTRODUCTION  Exports and imports play a crucial role in the economic stability of a country as they influence the amount of foreign exchange reserves.The need for foreign exchange is a vital factor in the monetary system of every country because it determines a nation's financial resilience.Many countries strive to bolster their foreign exchange reserves by promoting and increasing exports as one of the sources of foreign exchange.Exports are a crucial sector of the economy that plays a significant role in expanding markets between countries, potentially leading to the growth of various industries.This, in turn, can stimulate other sectors of the economy (Baldwin, 2005).The problem in this research is that the number of exported goods changes every time or season.That's why we need a mathematical model to predict the amount of exports in the future.
Time series analysis is an analysis using statistical techniques through the operation of models that use data from the past to predict the future.(Wei, W. W.S., 2006).STL is a method for decomposing a time series data into three components: seasonal, trend, and remainder.The definition of Loess is a non parametric regression method.This regression method has the advantage of high flexibility because the data automatically forms curve estimates that are not influenced by subjective factors (outliers) (Haritsah, 2015).In export data, there are often seasonal fluctuations caused by various factors, and STL can help effectively separate these seasonal components.

r
The application of Seasonal Trend Decomposition Using Loess for Export Forecasting by Economic Commodity Group in North Sumatra (Fahira Audri Yunisa) 62

RESEARCH METHODE
This research uses quantitative research.The form of data used in this research is in the form of time series which is secondary data, the acquisition of data used for this research is the total value of exports by economic goods group in North Sumatra and consists of 3 influential variables,namely barang modal (BM), bahan baku/penolong (BP), and barang konsumsi (BK).The data taken in this study are data on the number of exports (tonnes) in the period January 2018 to December 2022.The stages in the data analysis process in this study include: 1. Data description 2. Application of STL to each export data component 3. Applying seasonal, trend, and residual components to each component 4. Data is analysed descriptively to see any trends or seasonal patterns that may exist.5. Apply the STL method to the time series data using an appropriate library, such as STL plus or forecasting in R. Select appropriate parameters, such as LOESS windows for seasonal and trend components.6. Perform forecasting on each component of export data 7. Calculate MAE and MAPE values 8. Interpretation of results

CONCLUSION
From this research, the following conclusions were drawn: 1.This research uses the STL method which functions to decompose and forecast the value of exports by group of economic goods in sumatera utara 2. It can be seen from the forecasting results that the export value is a component that has a trend and seasonal pattern.3. From the forecasting results it can be concluded that the largest BM export value is 6357.6131(tons), the largest BP export value is 859804.0(tons) and the largest BP export value is 113157.64(tons).

3. 1 Figure 3 . 1 Figure 3 .
Figure 3.1 Chart of export data pattern January 2018 -December 2022 (tonnes) Figure 3.1 shows that the number of exports by economic goods group is unstable and has a seasonal pattern.

3. 2 Figure 3 . 2 Figure 3 . 5
Figure 3.2 Application of STL and forecasting on each export component in R studio From the application of STL, forecasting is obtained:

3. 3
Calculating MAE and MAPE values Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) are evaluation metrics used to measure the quality and accuracy of a forecasting model.Both provide information on how well the forecasting model matches the actual data.With the help of Microsoft Excel obtained: • MAE BM = 5416.38199MAPE BM = 1357.237• MAE BP = 187622.757MAPE BP = 38.21681• MAE BK = 34043.7192MAE BK = 120.3839