<![CDATA[In many cases, incomplete data is commonly found on both domestic and international websites. This is problematic for researchers who need complete data for their research, including in obtaining export values from the BPS Indonesia website. To address this issue, various prediction models can be used, including ARIMA (Auto Regressive Integrated Moving Average) which is a forecasting model based on statistics. With the completed library module, the Python language is now capable of running this model. ARIMA is a model that relies on try and error, so expertise is needed in determining its parameters. The problem of this research is to compare the manual ARIMA model with auto-ARIMA in Python using libraries that are available in this programming language. The purpose of this research is to get the best accuracy value of determining the parameters of manual and auto models in ARIMA. From the results of the research, it is concluded that the manually implemented ARIMA model performed better in MAE, MAPE and RMSE values compared to to the auto-ARIMA model, with values of 0.06 compared to 0.3, 0.006 compared to 0.03 and 0.07 compared to 0.4, respectively .]]>
Copyrights © 2024
