Multivariate LSTM versus Simple Baselines for FFB Yield Forecasting at Marihat Plantation: A Feasibility Pilot

Rendi Tajemir; Raden Aris Sugianto; Sri Lestari Rahayu

doi:10.61255/decoding.v4i2.1715

Authors

Rendi Tajemir Institut Teknologi Sawit Indonesia
Raden Aris Sugianto Institut Teknologi Sawit Indonesia
Sri Lestari Rahayu Institut Teknologi Sawit Indonesia

DOI:

https://doi.org/10.61255/decoding.v4i2.1715

Keywords:

Deep learning, Fresh Fruit Bunches (FFB), Long Short-Term Memory (LSTM), Production prediction, Time series

Abstract

Purpose – This study re-evaluates a multivariate LSTM for forecasting oil palm fresh fruit bunch (FFB) yield at Marihat Plantation after an audit identified a data-export bug in the original dataset. The data were rebuilt and assessed using a leakage-safe chronological design, naive and moving-average baselines, and seed-sensitivity checks.
Methods – Monthly FFB production, rainfall, and rainy-day records from plantation reports (2020–2023) were reconstructed and aggregated by planting-year cohort because block-level data were unavailable. Twelve-month windows across 11 cohorts yielded 240 sequences. With an 11-month purge gap, the chronological split comprised 10 training, 20 validation, and 10 test sequences. A two-layer LSTM (64 and 32 units; 20% dropout) was compared with five naive and moving-average baselines and retrained using 10 random seeds.
Findings – The LSTM achieved MAPE = 29.73% (MAE = 0.27; RMSE = 0.34 ton/ha) on the test set, outperforming all baselines (MAPE = 74–118%). Across 10 seeds, mean MAPE was 31.79% (SD = 4.59%; worst = 40.60%), remaining below every baseline. However, all test sequences represented December 2023 across 10 cohorts. The observed advantage therefore indicates feasibility, not forecasting superiority across time.
Research implications – The LSTM produced consistently lower errors than the baselines across cohorts for one target month, but the 10-sequence, single-month test set does not support general temporal claims. The prototype dashboard may support scenario exploration, although its 12-month forecasts depend on deterministic, unvalidated synthesized inputs.
Originality/value – Prompted by a data-quality audit documented in Supplementary S1, this study transparently rebuilds the evidence base and reframes the model as a feasibility pilot rather than a validated solution. Longer, gap-free records covering multiple test months and a validated multi-step procedure are required to establish a dependable advantage.

Abstract views: 0 ,

PDF downloads: 0

Downloads

Download data is not yet available.

References

T. F. Suardi, L. Sulistyowati, T. I. Noor, and I. Setiawan, “Analysis of the Sustainability Level of Smallholder Oil Palm Agribusiness in Labuhanbatu Regency, North Sumatra,” Agric., vol. 12, no. 9, 2022, doi: 10.3390/agriculture12091469.

N. Khan et al., “Prediction of Oil Palm Yield Using Machine Learning in the Perspective of Fluctuating Weather and Soil Moisture Conditions: Evaluation of a Generic Workflow,” Plants, vol. 11, no. 13, 2022, doi: 10.3390/plants11131697.

N. H. Setiawan and Zulkarnain, “Forecasting Palm Oil Production Using Long Short-Term Memory (Lstm) With Time Series Cross Validation (Tscv),” Int. J. Soc. Serv. Res., vol. 4, no. 05, pp. 1237–1251, 2024, doi: 10.46799/ijssr.v4i05.780.

A. Casolaro, V. Capone, G. Iannuzzo, and F. Camastra, “Deep Learning for Time Series Forecasting: Advances and Open Problems,” Inf., vol. 14, no. 11, 2023, doi: 10.3390/info14110598.

B. Lim and S. Zohren, “Time-series forecasting with deep learning: a survey,” Philos. Trans. R. Soc. A, vol. 379, no. 2194, p. 20200209, 2021, doi: 10.1098/rsta.2020.0209.

S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Comput., vol. 9, no. 8, pp. 1735–1780, Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.

M. Krichen and A. Mihoub, “Long Short-Term Memory Networks: A Comprehensive Survey,” AI, vol. 6, no. 9, pp. 1–21, 2025, doi: 10.3390/ai6090215.

I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016. [Online]. Available: https://www.deeplearningbook.org/

F. Ramadhan and M. Fachrie, “Implementasi Algoritma Long Short-Term Memory Pada Sistem Prediksi Hasil Panen Sawit,” J. Inform. Teknol. dan Sains, vol. 6, no. 4, pp. 937–944, Nov. 2024, doi: 10.51401/jinteks.v6i4.4876.

A. P. Putra and A. M. Harahap, “Implementasi Time Series Forecasting dengan Algoritma LSTM untuk Pemantauan dan Prediksi Produktivitas Kelapa Sawit Berdasarkan Hasil Panen,” JURIKOM (Jurnal Ris. Komputer), vol. 12, no. 2, pp. 46–56, Apr. 2025, doi: 10.30865/jurikom.v12i2.8495.

F. Husaini, I. Permana, M. Afdal, and F. N. Salisah, “Penerapan Algoritma Long Short-Term Memory untuk Prediksi Produksi Kelapa Sawit,” MALCOM Indones. J. Mach. Learn. Comput. Sci., vol. 4, no. 2, pp. 366–374, Feb. 2024, doi: 10.57152/malcom.v4i2.1187.

M. Syarovy et al., Prediction of Oil Palm Production Using Recurrent Neural Network Long Short-Term Memory (RNN-LSTM), vol. 1. Atlantis Press International BV, 2023. doi: 10.2991/978-94-6463-122-7_6.

K. Aqbar and R. A. Supomo, “Performance Analysis of LSTM and XGBoost Models Optimization in Forecasting Crude Palm Oil (CPO) Production at Palm Oil Mill (POM),” Int. J. Comput. Appl., vol. 185, no. 17, pp. 37–44, Jun. 2023, doi: 10.5120/ijca2023922890.

N. Khan, M. A. Kamaruddin, U. U. Sheikh, Y. Yusup, and M. P. Bakht, “Oil Palm and Machine Learning: Reviewing One Decade of Ideas, Innovations, Applications, and Gaps,” Agriculture, vol. 11, no. 9, p. 832, Aug. 2021, doi: 10.3390/agriculture11090832.

H. Hewamalage, K. Ackermann, and C. Bergmeir, “Forecast evaluation for data scientists: common pitfalls and best practices,” Data Min. Knowl. Discov., vol. 37, no. 2, pp. 788–832, Mar. 2023, doi: 10.1007/s10618-022-00894-5.

A. Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow (3rd Edition), O’Reilly Media, 2022. [Online]. Available: https://www.oreilly.com/library/view/hands-on-machine-learning/9781098125967/

N. I. A. Kader, U. K. Yusof, M. N. A. Khalid, and N. R. N. Husain, “A Review of Long Short-Term Memory Approach for Time Series Analysis and Forecasting,” in Proceedings of the 2nd International Conference on Emerging Technologies and Intelligent Systems (ICETIS 2022), M. A. Al-Sharafi, M. Al-Emran, M. N. Al-Kabi, and K. Shaalan, Eds., Cham: Springer, 2023, pp. 12–21. doi: 10.1007/978-3-031-20429-6_2.

J. Brownlee, Deep Learning for Time Series Forecasting: Predict the Future with MLPs, CNNs and LSTMs in Python. Machine Learning Mastery, 2018. [Online]. Available: https://machinelearningmastery.com/deep-learning-for-time-series-forecasting/

H. Dhake, Y. Kashyap, and P. Kosmopoulos, “Algorithms for Hyperparameter Tuning of LSTMs for Time Series Forecasting,” Remote Sens., vol. 15, no. 8, p. 2076, Apr. 2023, doi: 10.3390/rs15082076.

A. Pranolo et al., “Enhanced Multivariate Time Series Analysis Using LSTM: A Comparative Study of Min-Max and Z-Score Normalization Techniques,” Ilk. J. Ilm., vol. 16, no. 2, pp. 210–220, Aug. 2024, doi: 10.33096/ilkom.v16i2.2333.210-220.

R. J. Hyndman and G. Athanasopoulos, Forecasting: Principles and Practice, 3rd ed. Monash University, 2021. [Online]. Available: https://otexts.com/fpp3/

D. Koutsandreas, E. Spiliotis, F. Petropoulos, and V. Assimakopoulos, “On the selection of forecasting accuracy measures,” J. Oper. Res. Soc., vol. 73, no. 5, pp. 937–954, May 2022, doi: 10.1080/01605682.2021.1892464.

C. D. Lewis, Industrial and Business Forecasting Methods: A Practical Guide to Exponential Smoothing and Curve Fitting. London: Butterworth Scientific, 1982.

E. Strøm and O. E. Gundersen, “Performance metrics for multi-step forecasting measuring win-loss, seasonal variance and forecast stability: an empirical study,” Appl. Intell., vol. 54, no. 21, pp. 10490–10515, Nov. 2024, doi: 10.1007/s10489-024-05715-4.

Y. Baghoussi, C. Soares, and J. Mendes-Moreira, “Corrector LSTM: built-in training data correction for improved time-series forecasting,” Neural Comput. Appl., vol. 36, no. 26, pp. 16213–16231, Sep. 2024, doi: 10.1007/s00521-024-09962-x.

W. Wang, J. Shao, and H. Jumahong, “Fuzzy inference-based LSTM for long-term time series prediction,” Sci. Rep., vol. 13, no. 1, p. 20359, Nov. 2023, doi: 10.1038/s41598-023-47812-3.