Implementasi Hybrid ARIMA-GRU dengan Optimasi Optuna untuk Prediksi Penjualan Sepeda Listrik Studi Kasus PT.X

Irawan, Herlambang Awan (2026) Implementasi Hybrid ARIMA-GRU dengan Optimasi Optuna untuk Prediksi Penjualan Sepeda Listrik Studi Kasus PT.X. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) Skrispi INDO_organized111.pdf Download (1MB)
	Text (Bab 1) Skrispi Indo_Bab 1.pdf Download (253kB)
	Text (Bab 2) Skrispi Indo_Bab2.pdf Restricted to Repository staff only until 7 September 2029. Download (854kB)
	Text (Bab 3) Skrispi Indo_Bab 3.pdf Restricted to Repository staff only until 7 September 2029. Download (732kB)
	Text (Bab 4) Skrispi Indo_Bab 4.pdf Restricted to Repository staff only until 7 September 2029. Download (2MB)
	Text (Bab 5) Skrispi Indo_Bab 5.pdf Download (235kB)
	Text (Daftar Pustaka) Skrispi Indo_Daftar Pustaka.pdf Download (192kB)
	Text (Lampiran) Skrispi Indo_Lampiran.pdf Restricted to Repository staff only Download (244kB)

Abstract

Antelope electric bicycle sales at PT. X during the period from January 2021 to December 2024 exhibited fluctuating patterns with an average monthly growth trend of 6.16%, creating uncertainty in inventory planning, product distribution, and sales target management. This condition highlights the need for a forecasting model capable of capturing both linear and nonlinear patterns in time series data. The urgency of this study lies in implementing a hybrid approach that combines the strengths of statistical methods and deep learning to improve sales forecasting accuracy. This study employs the Autoregressive Integrated Moving Average (ARIMA) model, and a hybrid ARIMA-GRU model optimized using Optuna to forecast Antelope electric bicycle sales. The novelty of this research lies in applying Optuna-based GRU hyperparameter optimization within the hybrid ARIMA-GRU model for an electric bicycle sales forecasting case study, as well as implementing a Streamlit-based user interface that enables users to perform interactive data analysis and forecasting. The study utilizes 1,461 daily sales observations of Antelope E-Bikes collected from 2021 to 2024. The objectives of this research are to analyze the characteristics of the sales data, compare the performance of ARIMA, GRU, and hybrid ARIMA-GRU models with Optuna optimization, and develop a user-friendly interface that facilitates the utilization of forecasting results. The results indicate that the Optuna-optimized Hybrid ARIMA-GRU achieved the best performance, with an RMSE of 1.7398, an MAE of 1.4321, and a MAPE of 4.6585%, outperforming the Optuna-optimized ARIMA model, which obtained an RMSE of 3.2442, an MAE of 2.6432, and a MAPE of 8.7241%. Furthermore, this research produced a Streamlit-based application featuring data analysis, preprocessing, model evaluation, and 5, 7, 10 -days forecasting using a one-step recursive forecasting approach, thereby enabling users to access and utilize sales forecasting results more effectively.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Damaliana, Aviolla Terza	NIDN0002089402	aviolla.terza.sada@upnjatim.ac.id
Thesis advisor	Saputra, Wahyu Syaifullah Jauharis	NIDN0725088601	wahyu.s.j.saputra.if@upnjatim.ac.id

Subjects:

H Social Sciences > HA Statistics

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Herlambang Awan Irawan

Date Deposited:

07 Sep 2026 02:22

Last Modified:

07 Sep 2026 02:22

URI:

https://repository.upnjatim.ac.id/id/eprint/59989