Forecasting the Demand of a Specific Water Pump Product with a 3-day Lead Time
JOIV : International Journal on Informatics Visualization · 31.07.2026 · 10 (4) · 1810
DOI: 10.62527/joiv.10.4.4400 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOIV : International Journal on Informatics Visualization, online erschienen am 31.07.2026. Verfasst von Jonathan Albright Wiladi und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
This study examines 3-day demand forecasting for a specific water pump product at a retail agricultural tools store using a real-world dataset from 02nd December 2020 to 11th January 2024. It compares univariate time series forecasting using Autoregressive Integrated Moving Average (ARIMA), with feature-based forecasting, which includes 1 dependent variable, future sales, and 13 independent variables such as past sales, price, date, Consumer Price Index (CPI), daily operating hours, open unemployment rate, bank interest rate, rainfall rate, and Covid level. Feature-based models were implemented using Random Forest, Gradient Boosting, and Support Vector Regressor (SVR). Gradient Boosting achieved the best performance numerically. However, a paired t-test at the 5% significance level shows no significant differences in Root Mean Squared Error (RMSE) and Root Mean Squared Scaled Error (RMSSE) among ARIMA, Gradient Boosting, and Random Forest. SVR had significantly higher RMSE and RMSSE than Gradient Boosting and Random Forest. Still, the difference from ARIMA was not statistically significant—no significant differences in Mean Absolute Error (MAE) between the 4 models. Feature importance analysis conducted on Gradient Boosting shows that CPI is the most contributing factor, while rainfall rate has the lowest importance. The result shows that a simulator supports demand prediction. Overall, feature-based forecasting offers modest improvements over univariate time series. This study contributes to demand forecasting for retail agricultural tools, identifies key predictive features, and demonstrates practical implementation. Future studies could evaluate multiple products, include broader variables, and use additional models to improve forecasting performance further.
Zitieren
APA
Wiladi, J. A., & Mauritsius, T. (2026). Forecasting the Demand of a Specific Water Pump Product with a 3-day Lead Time. JOIV : International Journal on Informatics Visualization, 10(4), 1810. https://doi.org/10.62527/joiv.10.4.4400
Vancouver
Wiladi JA, Mauritsius T. Forecasting the Demand of a Specific Water Pump Product with a 3-day Lead Time. JOIV : International Journal on Informatics Visualization. 2026;10(4):1810. doi:10.62527/joiv.10.4.4400
BibTeX
@article{Wiladi2026,
author = {Wiladi, Jonathan Albright and Mauritsius, Tuga},
title = {Forecasting the Demand of a Specific Water Pump Product with a 3-day Lead Time},
journal = {JOIV : International Journal on Informatics Visualization},
year = {2026},
volume = {10},
number = {4},
pages = {1810},
doi = {10.62527/joiv.10.4.4400}
}
EUVIMED