Comparison in Predicting the Short-Term Using the Sarima, Dsarima and Tsarima Methods

Abstract

The flow of data and information is growing rapidly and rapidly in various sizes and means which is called Big Data. In the face of a change for the better in the future, a careful analysis and design of a data processing system is needed, in which a predictive framework can formulate the right policy to be one of the efforts to make a good decision. This is one of the appropriate Big Data processing efforts, which can be realized through one of the methods, namely prediction or forecasting is an effort to predict future values or trends as a reference for analyzing conditions in the past. One example of Big Data in the City of Balikpapan, namely the temperature within 2 meters obtained from the NASA satellite published on the website power.larc.nasa.gov. One of the methods used in this research is the ARIMA method and it is developed according to the data used. Based on the data to be used, namely temperature data within a distance of 2 meters in the city of Balikpapan, the development of data processing is carried out to pay attention to three seasonal patterns or the so-called Triple Seasonal ARIMA model. In this study, it can be seen how to build the Triple Seasonal ARIMA model and comparison with alternative models, namely Seasonal ARIMA and Double Seasonal ARIMA, and can see how the results of the Triple Seasonal ARIMA model accuracy when compared with alternative models. The method used in this study is the Seasonal ARIMA, Double Seasonal ARIMA and Triple Seasonal ARIMA methods. The results obtained in this study obtained a comparison of methods in making predictions with a specified time span, the results obtained from the Seasonal ARIMA model that it was very good at predicting a time span of 2 weeks, Double Seasonal ARIMA for a period of 1 month, Double Seasonal ARIMA for a period of 3 months, and Triple Seasonal ARIMA for a period of 6 months.

Triple-Seasonal ARIMA Untuk Peramalan Data Konsumsi Beban Listrik

“…Pola data musiman merupakan fluktuasi data yang berulang secara berkala akibat faktor musiman, seperti cuaca, hari, minggu, bulan, maupun tahun [2]. Sehingga, saat terdapat pola musiman, dapat menggunakan metode Seasonal ARIMA (SARIMA) yang merupakan pengembangan dari metode ARIMA.…”

“…Jika data mengalami pola musiman ganda berulang seperti hari dan tahun, maka dapat menggunakan metode Double SARIMA yang merupakan pengembangan dari metode SARIMA. Beberapa metode peramalan dengan tingkat keberhasilan yang bervariasi telah diimplementasikan untuk peramalan beban listrik termasuk pada penelitian Syalam, dkk [2][3][4]. Selanjutnya, terdapat juga pengembangan dari metode Double SARIMA yaitu Triple SARIMA yang berkaitan dengan data seasonal berulang seperti hari, bulan, dan tahun.…”