Please use this identifier to cite or link to this item: https://repositori.uma.ac.id/handle/123456789/30269
Title: Peramalan Jumlah Pengunjung Website Menggunakan Algoritma Prophet dengan Optimasi Slime Mould Algorithm
Other Titles: Forecasting Website Visitor Numbers Using the Prophet Algorithm with Slime Mould Algorithm Optimization
Authors: Br Nainggolan, Adelina
metadata.dc.contributor.advisor: Lubis, Andre Hasudungan
Keywords: Peramalan deret waktu;Prophet;Slime Mould Algorithm;Optimasi hyperparameter;Metaheuristik;Time series forecasting;Hyperparameter optimization
Issue Date: Feb-2026
Publisher: Universitas Medan Area
Series/Report no.: NPM;228160093
Abstract: Peramalan deret waktu merupakan permasalahan penting dalam analisis data karena berkaitan dengan kemampuan model dalam menangkap pola tren dan musiman secara akurat. Algoritma Prophet merupakan salah satu metode peramalan deret waktu yang banyak digunakan karena kemampuannya dalam memodelkan komponen tren dan musiman. Namun, kinerja algoritma Prophet sangat dipengaruhi oleh pemilihan hyperparameter, sehingga penggunaan parameter default sering kali menghasilkan tingkat akurasi yang belum optimal. Oleh karena itu, penelitian ini bertujuan untuk meningkatkan kinerja model Prophet melalui optimasi hyperparameter menggunakan Slime Mould Algorithm (SMA). Penelitian ini menggunakan data deret waktu sebagai studi kasus pengujian model. Tahapan penelitian meliputi prapemrosesan data, proses optimasi hyperparameter Prophet menggunakan SMA, pembangunan model peramalan, serta evaluasi kinerja model. Hyperparameter yang dioptimasi meliputi changepoint prior scale dan seasonality prior scale. Evaluasi kinerja model dilakukan dengan membandingkan nilai kesalahan menggunakan metrik Root Mean Squared Error (RMSE) dan Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan bahwa optimasi hyperparameter menggunakan SMA mampu meningkatkan kinerja model Prophet. Nilai RMSE berhasil diturunkan dari 399,7137 menjadi 338,0901, sedangkan nilai MAPE menurun dari 13,5% menjadi 11,47%. Selain itu, hasil peramalan menunjukkan bahwa model Prophet yang telah dioptimasi lebih adaptif terhadap perubahan tren dan fluktuasi pada data deret waktu. Dengan demikian, SMA terbukti efektif sebagai metode optimasi hyperparameter dalam meningkatkan akurasi dan stabilitas model peramalan Prophet. Time series forecasting is a crucial issue in data analysis because it is related to the model's ability to accurately capture trend and seasonal patterns. The Prophet algorithm is a widely used time series forecasting method due to its ability to model trend and seasonal components. However, the performance of the Prophet algorithm is strongly influenced by hyperparameter selection, so the use of default parameters often results in a suboptimal level of accuracy. Therefore, this study aims to improve the performance of the Prophet model through hyperparameter optimization using the Slime Mold Algorithm (SMA). This study uses time series data as a case study for model testing. The research stages include data preprocessing, the Prophet hyperparameter optimization process using SMA, forecasting model development, and model performance evaluation. The optimized hyperparameters include the changepoint prior scale and the seasonality prior scale. Model performance evaluation is carried out by comparing error values using the Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) metrics. The results show that hyperparameter optimization using SMA can improve the performance of the Prophet model. The RMSE value was successfully reduced from 399.7137 to 338.0901, while the MAPE value decreased from 13.5% to 11.47%. Furthermore, the forecasting results show that the optimized Prophet model is more adaptive to changing trends and fluctuations in time series data. Thus, the Slime Mold Algorithm has proven effective as a hyperparameter optimization method in improving the accuracy and stability of the Prophet forecasting model.
Description: 53 Halaman
URI: https://repositori.uma.ac.id/handle/123456789/30269
Appears in Collections:SP - Informatic Engineering

Files in This Item:
File Description SizeFormat 
228160093 - Adelina Br Nainggolan - Fulltext.pdfCover, Abstract, Chapter I, II, III, V, Bibliography803.59 kBAdobe PDFView/Open
228160093 - Adelina Br Nainggolan - Chapter IV.pdf
  Restricted Access
Chapter IV453.36 kBAdobe PDFView/Open Request a copy


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.