High performance time series models using auto autoregressive integrated moving average

Abstract

Forecasting techniques have received considerable interest from both researchers and academics because of the unique characteristics of businesses and their influence on several areas of the economy. Most academics utilize the autoregressive integrated mov ing average (ARIMA) approach to forecasting the future. However, researchers face challenges, such as analyzing the data and selecting the appropriate ARIMA parameters, especially with large datasets. This study investigates the use of the automatic ARIMA (Auto ARIMA) function for forecasting Brent oil prices. It demonstrates the benefits of using Auto ARIMA over ARIMA for determining the appropriate ARIMA parameters based on measures such as root mean square error ( RMSE ) , mean absolute error ( MAE ) , and aka ike information criterion ( AIC ) without requiring the attention of an expert data scientist as it bypasses several steps needed for manual ARIMA. Auto ARIMA produced an RMSE of 12.5539 and an AIC of 1877.224, which are comparable to the values resulting fr om the manual ARIMA with the help of expert data scientists; thus, it saves analysis time and offers the best model result.