Using a novel clustered 3D-CNN model for improving crop future price prediction

Abstract

Many research studies used statistics to predict food production, distribution and price trends. Researchers used statistical inference for discovering the relationships of data to build predictive model. However, crop production and its price trend do not only depend on ecosystems, molecular biology, precision agriculture, veterinary science, animal genes, and technology, but also depend on the environmental change and economic factors. Most importantly, the crop price trend is in non-stationary pattern and is influenced by multiple dimensional factors that the traditional techniques of time series forecasting, such as ARIMA, cannot perform well in prediction. Since CNN model can cope with non-stationary data and learn non-linearity by adjusting the model parameters, it can overcome the limitations of the traditional statistical methods in prediction. Therefore, the aims of this research are to conduct a review to identify a more complete factors that may influence crop production and price changes, and to propose a novel Clustered 3D-CNN model for predicting crop future price. The experiments to compare the performance of our proposed model and ARIMA model were done. The average results found that our proposed Clustered 3D-CNN model (MAPE = 0.083, RMSE = 40.39, MAE = 32.31) outperforms the ARIMA model (MAPE = 0.108, RMSE = 59.95, MAE = 46.35). The 3D-CNN model helps decision makers to better predict crop price trend, and to develop a strategic plan for selecting trading partners to reduce the cost and for solving food insecurity problem.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 54 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.