Onion wholesale price forecasting: A comprehensive study of traditional time series and AI-based approaches

Jaehong Park; Seungjee Hong

doi:10.7744/kjoas.530305

All Issue

2026 Vol.53, Issue 3 Preview Page Next Page

Management & Economics

Onion wholesale price forecasting: A comprehensive study of traditional time series and AI-based approaches

1 September 2026. pp. 341-358

PDF XML

Abstract

Agricultural price forecasting studies increasingly favor machine learning, deep learning, and hybrid models over traditional statistical approaches, yet few studies test whether this added complexity is justified under realistic data constraints. This study compares seven forecasting methodologies for onion wholesale price using daily data from the Garak wholesale market (June 2023 - December 2025, 788 observations): traditional time series models (autoregressive integrated moving average [ARIMA], seasonal autoregressive integrated moving average [SARIMA]), machine learning (extreme gradient boosting [XGBoost]), deep learning (long short-term memory [LSTM], gated recurrent unit [GRU]), and hybrid models (ARIMA-LSTM, variational mode decomposition [VMD]-LSTM). Models were evaluated via leakage-free rolling one-day-ahead walk-forward forecasts using mean absolute error (MAE), root mean squared error (RMSE), mean absolute percentage error (MAPE), coefficient of determination (R²), and directional accuracy, with Diebold-Mariano tests, and further compared across forecast horizons (1, 7, 14, and 30 days) and seasons. Contrary to common expectations, the simplest models performed competitively: ARIMA (MAE 17.39 won/kg, R² = 0.518) and GRU (MAE 17.67 won/kg, R² = 0.564) achieved the best accuracy, SARIMA showed no significant improvement over ARIMA (p = 0.183), and VMD-LSTM was significantly outperformed by the simpler ARIMA-LSTM (p < 0.001). XGBoost’s one-day error exceeded ARIMA’s and SARIMA’s, and deteriorated most severely among all models as the horizon lengthened due to recursive error accumulation. These results align with evidence that machine learning needs large training samples to outperform simple statistical benchmarks, and that reported gains from VMD hybrids may partly reflect information leakage, indicating that complexity does not guarantee performance under limited data. The findings offer practical guidance for selecting forecasting models in data-constrained agricultural price settings.

Keywords

hybrid forecasting models

information leakage

onion wholesale price

small-sample forecasting

References

Aiyegbeni G, Li Y. 2024. A comparative analysis of LSTM, ARIMA, XGBoost algorithms in predicting stock price direction. Engineering and Technology Journal 9:4978-4986.

10.47191/etj/v9i08.50

Aritonang PK, Wiryono SK, Faturohman T. 2024. A comparative analysis of deep learning and traditional statistics for stock price and return forecasting. Global Business & Finance Review 29:1-17.

10.17549/gbfr.2024.29.8.1

Blanchet J, Hernandez F, Nguyen VA, Pelger M, Zhang X. 2022. Bayesian imputation with optimal look-ahead-bias and variance tradeoff. arXiv preprint arXiv:2202.00871.

10.2139/ssrn.4024810

Box GEP, Jenkins GM. 1976. Time series analysis: Forecasting and control. Holden-Day, San Francisco, CA.

Brignoli PL, Varacca A, Gardebroek C, Sckokai P. 2024. Machine learning to predict grains futures prices. Agricultural Economics 55:479-497.

10.1111/agec.12828

Cerqueira V, Torgo L, Soares C. 2022. A case study comparing machine learning with statistical methods for time series forecasting: Size matters. Journal of Intelligent Information Systems 59:415-433.

10.1007/s10844-022-00713-9

Chandran KP, Pandey NK. 2007. Potato price forecasting using seasonal ARIMA approach. Potato Journal 34:137-138.

Chen T, Guestrin C. 2016. XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. pp. 785-794. New York, NY: ACM.

10.1145/2939672.2939785

Cho K, van Merriënboer B, Gulcehre C, Bahdanau D, Bougares F, Schwenk H, Bengio Y. 2014. Learning phrase representations using RNN encoder–decoder for statistical machine translation. In: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. pp. 1724-1734. Stroudsburg, PA: ACL.

10.3115/v1/D14-1179

Dauda TO, Tiamiyu-Ibrahim S. 2021. Autoregressive integration moving average (ARIMA) model for prices of selected grains in the South West Nigeria. Agricultural Science Digest 41:35-41.

10.18805/ag.D-239

Diebold FX, Mariano RS. 1995. Comparing predictive accuracy. Journal of Business & Economic Statistics 13:253-263.

10.1080/07350015.1995.10524599

Dragomiretskiy K, Zosso D. 2014. Variational mode decomposition. IEEE Transactions on Signal Processing 62:531-544.

10.1109/TSP.2013.2288675

Feng W, Tao R, Cartlidge J, Zheng J. 2026. VMDNet: Temporal leakage-free variational mode decomposition for electricity demand forecasting. In: Proceedings of the 34th European Signal Processing Conference (EUSIPCO 2026). Belgium: EURASIP.

Hochreiter S, Schmidhuber J. 1997. Long short-term memory. Neural Computation 9:1735-1780.

10.1162/neco.1997.9.8.1735

Huang X, Zhuang X, Tian F, Niu Z, Chen Y, Zhou Q, Yuan C. 2025. A hybrid ARIMA-LSTM-XGBoost model with linear regression stacking for transformer oil temperature prediction. Energies 18:1432.

10.3390/en18061432

Jang SH, Chun HJ, Cho IH, Kim DH. 2017. A study on cabbage wholesale price forecasting model using unstructured agricultural meteorological data. Journal of the Korean Data & Information Science Society 28:617-624. [in Korean]

10.7465/JKDI.2017.28.3.617

Jung DH, Cho YY. 2022. A prediction model for Hallabong tangor product prices using LSTM (long short-term memory) network. Horticultural Science and Technology 40:571-577. [in Korean]

10.7235/HORT.20220051

KAMIS (Korea Agricultural Marketing Information Service). 2026. Price information. Accessed in https://www.kamis.or.kr/customer/info/market/period.do on 15 January 2026. [in Korean]

Kim BS. 2005. A comparison on forecasting performance of the application models for forecasting of vegetable prices. The Korean Journal of Agricultural Economics 46:89-113. [in Korean]

Little RJA, Rubin DB. 2019. Statistical analysis with missing data (3rd). John Wiley & Sons, Hoboken, NJ.

10.1002/9781119482260

Liu T, Ma X, Li S, Li X, Zhang C. 2022. A stock price prediction method based on meta-learning and variational mode decomposition. Knowledge-Based Systems 252:109324.

10.1016/j.knosys.2022.109324

Makridakis S, Spiliotis E, Assimakopoulos V. 2018. Statistical and machine learning forecasting methods: Concerns and ways forward. PLOS ONE 13:e0194889.

10.1371/journal.pone.019488929584784PMC5870978

Nichani R, Gasmi L, Laiche N, Kabou S. 2024. Optimizing financial time series predictions with hybrid ARIMA, LSTM, and XGBoost models. Studies in Engineering and Exact Sciences 5:e11188.

10.54021/seesv5n2-582

Shin SC, Lee MS, Kim ME, Byun JH, Kim DH. 2026. Chapter 9. Condiment vegetables: Supply-demand trends and outlook. Agricultural Outlook 2026. Report No. E04-2026. pp. 483-484. Korea Rural Economic Institute. [in Korean]

Tashman LJ. 2000. Out-of-sample tests of forecasting accuracy: An analysis and review. International Journal of Forecasting 16:437-450.

10.1016/S0169-2070(00)00065-0

Tsay RS. 2010. Analysis of financial time series (3rd). John Wiley & Sons, Hoboken, NJ.

10.1002/9780470644560

Xia C. 2024. Comparative analysis of ARIMA and LSTM models for agricultural product price forecasting. Highlights in Science, Engineering and Technology 85:1032-1040.

10.54097/8q6nx369

Yang Z, Zhang R, Yin Z, Zhu Y. 2022. Pork price prediction using LSTM model: Based on a new dataset. In: Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology. pp. 423-429. China: ICPDI.

10.5220/0011738900003607

Zhang GP. 2003. Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing 50:159-175.

10.1016/S0925-2312(01)00702-0

Information

Publisher :Institute of Agricultural Science, Chungnam National University
Publisher(Ko) :충남대학교 농업과학연구소
Journal Title :Korean Journal of Agricultural Science
Journal Title(Ko) :농업과학연구
Volume : 53
No :3
Pages :341-358
Received Date : 2026-07-20
Revised Date : 2026-08-02
Accepted Date : 2026-08-02
DOI :https://doi.org/10.7744/kjoas.530305

Korean Journal of Agricultural Science ISSN:2466-2402(Print) 2466-2410(Online) 농업과학연구

All Issue