Analisis Peramalan Menggunakan Metode SARIMAX, ANN, dan Hibrida SARIMAX-ANN paAnalisis peramalan menggunakan metode SARIMAX, ANN, dan Hibrida SARIMAX-ANN pada data time seriesda Data Time Series

Fatona, Nabila Ridha (2025) Analisis Peramalan Menggunakan Metode SARIMAX, ANN, dan Hibrida SARIMAX-ANN paAnalisis peramalan menggunakan metode SARIMAX, ANN, dan Hibrida SARIMAX-ANN pada data time seriesda Data Time Series. Sarjana thesis, UIN Sunan Gunung Djati Bandung.

[img]
Preview
Text (COVER)
1_cover.pdf

Download (235kB) | Preview
[img]
Preview
Text (ABSTRAK)
2_abstrak.pdf

Download (285kB) | Preview
[img]
Preview
Text (KETERANGAN BEBAS PLAGIARISM)
3_skbebasplagiarism.pdf

Download (65kB) | Preview
[img]
Preview
Text (DAFTAR ISI)
4_daftarisi.pdf

Download (269kB) | Preview
[img]
Preview
Text (BAB I)
5_bab1.pdf

Download (300kB) | Preview
[img] Text (BAB II)
6_bab2.pdf
Restricted to Registered users only

Download (737kB) | Request a copy
[img] Text (BAB III)
7_bab3.pdf
Restricted to Registered users only

Download (584kB) | Request a copy
[img] Text (BAB IV)
8_bab4.pdf
Restricted to Registered users only

Download (621kB) | Request a copy
[img] Text (BAB V)
9_bab5.pdf
Restricted to Registered users only

Download (283kB) | Request a copy
[img] Text (DAFTAR PUSTAKA)
10_daftarpustaka.pdf
Restricted to Registered users only

Download (247kB) | Request a copy
[img] Text (LAMPIRAN)
11_lampiran.pdf
Restricted to Repository staff only

Download (592kB) | Request a copy

Abstract

INDONESIA: Peramalan time series merupakan salah satu metode dalam pengambilan keputusan, khususnya dalam sektor transportasi yang sangat bergantung pada pola musiman dan berbagai faktor eksternal. Jumlah penumpang kereta api non-Jabodetabek yang fluktuatif setiap bulannya menjadi tantangan dalam proses peramalan, terlebih karena dipengaruhi oleh hari libur. Penelitian ini bertujuan untuk menganalisis dan membandingkan kinerja tiga pendekatan peramalan, yaitu model Seasonal Autoregressive Integrated Moving Average Exogenous (SARIMAX), Artificial Neural Network (ANN), dan model hibrida SARIMAX-ANN. Metode penelitian ini diawali dengan transformasi dan differencing terhadap data deret waktu agar menjadi stasioner, diikuti dengan pencarian model SARIMAX. Residual dari model SARIMAX kemudian digunakan sebagai target pelatihan model ANN, dengan input berupa lag residual dan variabel eksogen (jumlah hari libur). Model hibrida dibentuk dengan menjumlahkan hasil prediksi SARIMAX dan prediksi residual dari ANN. Hasil evaluasi menunjukkan bahwa model SARIMAX memberikan hasil dengan nilai RMSE sebesar 7144.42 dan MAE sebesar 6595.72, model ANN memberikan hasil dengan nilai RMSE 911.1169 dan MAE sebesar 682.0064, dan model hibrida SARIMAX-ANN memberikan performa peramalan terbaik dengan nilai RMSE sebesar 645.89 dan MAE sebesar 475.86, jauh lebih baik dibandingkan SARIMAX maupun ANN tunggal. Dengan demikian, pendekatan hybrid terbukti efektif dalam meningkatkan akurasi peramalan time series yang kompleks dan dipengaruhi oleh faktor eksternal. ENGLISH: Time series forecasting is one of the methods in decision making, especially in the transportation sector which is highly dependent on seasonal patterns and various external factors. The number of non-Jabodetabek train passengers that fluctuates every month is a challenge in the forecasting process, especially because it is influenced by holidays. This study aims to analyze and compare the performance of three forecasting approaches, namely the Seasonal Autoregressive Integrated Moving Average Exogenous (SARIMAX) model, Artificial Neural Network (ANN), and SARIMAX-ANN hybrid model. The method of this research begins with the transformation and differencing of time series data to make it stationary, followed by the search for the SARIMAX model. The residuals from the SARIMAX model are then used as the training target for the ANN model, with inputs in the form of lag residuals and exogenous variables (number of holidays). The hybrid model is formed by summing the SARIMAX prediction results and the residual prediction from the ANN. The evaluation results show that the SARIMAX model provides results with an RMSE value of 7144.42 and an MAE of 6595.72, the ANN model provides results with an RMSE value of 911.1169 and an MAE of 682.0064, and the SARIMAX hybrid model ANN hybrid model provides the best forecasting performance with an RMSE value of 645.89 and an MAE of 475.86, significantly better than either the SARIMAX or ANN models alone. Thus, the hybrid approach has proven effective in improving the accuracy of forecasting complex time series influenced by external factors.

Item Type: Thesis (Sarjana)
Uncontrolled Keywords: Peramalan; SARIMAX; Artificial Neural Network; Hibrida SARIMAX-ANN; Time Series
Subjects: Mathematics > Data Processing and Analysis of Mathematics
Applied mathematics > Statistical Mathematics
Applied mathematics > Descriptive Statistical Mathematics
Divisions: Fakultas Sains dan Teknologi > Program Studi Matematika
Depositing User: Nabila Ridha Fatona
Date Deposited: 25 Aug 2025 04:29
Last Modified: 25 Aug 2025 04:29
URI: https://digilib.uinsgd.ac.id/id/eprint/115911

Actions (login required)

View Item View Item