FORECASTNG BANK TRANSACTIONS WITH TIME SERIES MODELS: A COMPARATIVE ANALYSIS OF ARIMA, SARIMA AND SARIMAX
Elmi Əsərlər · 2025
DOI: 10.61413/irwl4767 ↗Kurz gefasst
Fachartikel in der Zeitschrift Elmi Əsərlər, 2025. Verfasst von Musa Elxan Rahimov und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The article applies ARIMA, SARIMA and SARIMAX models to forecast transaction volumes in the banking sector. The ARIMA model adjusts only for trend and stationarity effects, SARIMA also takes into consideration seasonal effects, and SARIMAX additionally increases the accuracy of the forecast by including the exogenous feature "Transaction Growth". The "Transaction Growth" feature is created using feature engineering and reflects changes in transaction volumes as a percentage value. The purpose of this feature is to improve the performance of models by providing deeper analysis of data. Different performance metrics such as RMSE, MAE, R², MAPE, sMAPE and MBE were applied to evaluate the performance of the models. The results obtained from the SARIMAX model indicate that this model provides the highest accuracy in comparison with other models: RMSE (22.61), MAE (17.48), R² Score (0.9125), MAPE (4.05%), sMAPE (4.04%) and MBE (-0.33). This study shows that the use of the "Transaction Growth" feature created through feature engineering in the models significantly increases the accuracy of time series forecasts. The advantage of the SARIMAX model in predicting transaction volumes in the banking sector is the most noticeable benefit of the study.
Zitieren
APA
Rahimov, M. E., & Namazov, M. B. (2025). FORECASTNG BANK TRANSACTIONS WITH TIME SERIES MODELS: A COMPARATIVE ANALYSIS OF ARIMA, SARIMA AND SARIMAX. Elmi Əsərlər. https://doi.org/10.61413/irwl4767
Vancouver
Rahimov ME, Namazov MB. FORECASTNG BANK TRANSACTIONS WITH TIME SERIES MODELS: A COMPARATIVE ANALYSIS OF ARIMA, SARIMA AND SARIMAX. Elmi Əsərlər. 2025. doi:10.61413/irwl4767
BibTeX
@article{Rahimov2025,
author = {Rahimov, Musa Elxan and Namazov, Manafaddin Bashir},
title = {FORECASTNG BANK TRANSACTIONS WITH TIME SERIES MODELS: A COMPARATIVE ANALYSIS OF ARIMA, SARIMA AND SARIMAX},
journal = {Elmi Əsərlər},
year = {2025},
doi = {10.61413/irwl4767}
}
EUVIMED