Tashreef Muhammad, Tahsin Ahmed, Meherun Farzana, Md. Mahmudul Hasan, Abrar Eyasir, Md. Emon Khan, Mahafuzul Islam Shawon, Ferdous Mondol, Mahmudul Hasan, Muhammad Ibrahim
arXiv 27 Mar 2026 · Machine Learning
arXiv:2604.06227 · PDF · DOI · OpenAlex · Extracted main text
Accurate short-term forecasting of agricultural commodity prices is critical for food security planning and smallholder income stabilisation in developing economies, yet machine-learning-ready datasets for this purpose remain scarce in South Asia. This paper makes two contributions. First, we introduce AgriPriceBD, a benchmark dataset of 1,779 daily retail mid-prices for five Bangladeshi commodities - garlic, chickpea, green chilli, cucumber, and sweet pumpkin - spanning July 2020 to June 2025, extracted from government reports via an LLM-assisted digitisation pipeline. Second, we evaluate seven forecasting approaches spanning classical models - naïve persistence, SARIMA, and Prophet - and deep learning architectures - BiLSTM, Transformer, Time2Vec-enhanced Transformer, and Informer - with Diebold-Mariano statistical significance tests. Commodity price forecastability is fundamentally heterogeneous: naïve persistence dominates on near-random-walk commodities. Time2Vec temporal encoding provides no statistically significant advantage over fixed sinusoidal encoding and causes catastrophic degradation on green chilli (+146.1% MAE, p<0.001). Prophet fails systematically, attributable to discrete step-function price dynamics incompatible with its smooth decomposition assumptions. Informer produces erratic predictions (variance up to 50x ground-truth), confirming sparse-attention Transformers require substantially larger training sets than small agricultural datasets provide. All code, models, and data are released publicly to support replication and future forecasting research on agricultural commodity markets in Bangladesh and similar developing economies.
appendix boundary found by appendix_command · 86% of the source is main text. Read the extracted text to check this.
The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.
| Reference | Intensity | Mentions | Sections | Main text | |
|---|---|---|---|---|---|
| 1 | Hyndman, Rob J and Athanasopoulos, George (2018) Forecasting: principles and practice | 1.000 | 5 | 3 | 100% |
| 2 | Imran, Abdullah Al and Wahid, Zaman and Prova, Alpana Akhi and Hanna… (2022) Harnessing the meteorological effect for predicting the retail price of rice in Bangladesh | 1.000 | 5 | 3 | 100% |
| 3 | Zhou, Haoyi and Zhang, Shanghang and Peng, Jieqi and Zhang, Shuai an… (2021) Informer: Beyond efficient transformer for long sequence time-series forecasting | 0.843 | 3 | 3 | 100% |
| 4 | Hassan, MF and Islam, MA and Imam, MF and Sayem, SM (2013) Forecasting wholesale price of coarse rice in Bangladesh: A seasonal autoregressive integrated moving average approach | 0.811 | 4 | 2 | 100% |
| 5 | Sari, Murat and Duran, Serbay and Kutlu, Huseyin and Guloglu, Bulent… (2024) Various optimized machine learning techniques to predict agricultural commodity prices | 0.811 | 4 | 2 | 100% |
| 6 | Hasan, Md Mehedi and Zahara, Muslima Tuz and Sykot, Md Mahamudunnobi… (2020) Ascertaining the fluctuation of rice price in Bangladesh using machine learning approach self | 0.737 | 3 | 2 | 100% |
| 7 | Kazemi, Seyed Mehran and Goel, Rishab and Eghbali, Sepehr and Ramana… (2019) Time2vec: Learning a vector representation of time | 0.737 | 3 | 2 | 100% |
| 8 | Manogna, RL and Dharmaji, Vijay and Sarang, S (2025) Enhancing agricultural commodity price forecasting with deep learning | 0.737 | 3 | 2 | 100% |
| 9 | Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Ja… (2017) Attention is all you need | 0.737 | 3 | 2 | 100% |
| 10 | Dasari, Suresh Babu and Para, Hemanth Suresh and Chaduvula, Dheeraz (2025) Price Forecasting for Vegetables using SARIMA-LSTM and Multitask Learning | 0.644 | 4 | 1 | 100% |
Showing the top 10 of 32 scored citations.