Comparison of Exponential Smoothing Methods for Forecasting Marine Fish Production in Pekalongan Waters, Central Java

Abstract

Pekalongan waters, a part of the Java Sea, has potency to develop marine fisheries sector to increase regional income and community livelihoods. The fluctuation of marine fish production every year requires serious attention in planning and policy strategies for the utilization of the fishery resources. Time series fish production data can be used to predict fish production in the following years through the forecasting process. The data used in this study is fish production data from Pekalongan Fishing Port, Central Java, from January 2011 to December 2020. The method used is data exponential smoothing by comparing three exponential smoothing methods consisting of single/simple exponential smoothing, double exponential smoothing and Holt-Winters’ exponential smoothing. The criterion that used to measure the forecasting performance is the mean absolute percentage error (MAPE) value. The smaller MAPE value shows the better the forecasting result. The smallest MAPE value is obtained by finding the optimal smoothing constant value which is usually calculated using the trial and error method. However, in this study, the constant value was calculated using the add-in solver approach in Microsoft Excel. The forecasting results obtained show that forecasting using the Holt Winter Exponential Smoothing method is reasonable with a MAPE value of 37.878.

Effectiveness of Single and Double Exponential Smoothing: SES, ARRSES and Holt’s Linear for Time Series Data Prediction with Trend and Non-seasonal Characteristic (Covid-19 Vaccinate Case)

“…Selain penelitian tersebut di atas, metode exponential smoothing juga digunakan dalam machine learning, dengan cakupan data yang sangat besar (big data) oleh Shastri dkk [11] dengan hasil bahwa metode exponential smoothing dapat digunakan untuk memprediksi lima tahun ke depan dengan cakupan data historis yang digunakan adalah tiga puluh lima tahun. Pamungkas dkk [8] meneliti mengenai komparasi metode exponential smoothing dalam bidang perikanan dan kelautan, yang mana hasil dari penelitiannya metode terbaik adalah exponential smoothing jenis Holt's yang memperoleh nilai peramalan dengan kesalahan 37,878%. Penelitian Kahraman dan Akay [5] membandingkan metode exponential smoothing untuk kasus data harga logam diperoleh hasil bahwa model terbaik untuk memprediksi untuk alumunium, tembaga, timah dan besi adalah damped trend, model Holt's untuk harga logam nikel dan seng, dan model Brown untuk harga logam timah.…”

NFT Investments Analysis: A Strategic Approach with Ranking Insights and Sales Forecasting System for Informed Decision-Making

“…In several articles, researchers have compared and examined various exponential smoothing techniques, including Holt's Double Exponential Smoothing (DES) and Triple Exponential Smoothing (TES). The approaches' efficacy varies based on the data's parameters and the problem that has to be solved (Siregar et al, 2017); (Pamungkas et al, 2021).…”

“…Single exponential smoothing, or SES, is a useful technique for univariate time series data that does not exhibit seasonal patterns or trends. This approach is usually applied when the data exhibits a steadily changing average but no discernible trend or seasonal pattern (Pamungkas et al, 2021).…”

Komparasi Metode Regresi Linier, Exponential Smoothing dan ARIMA Pada Peramalan Volume Ekspor Minyak Kelapa Sawit di Indonesia

“…Ratarata data mengalami perubahan secara perlahan dari waktu ke waktu. Rumus yang digunakan pada metode ini adalah sebagai berikut [7]:…”