ABSTRACT

This research specifically addresses the prediction of dairy product demand (DPD). Since dairy products have a short consumption period, it is important to have accurate information about their future demand. The main contribution of this research is to provide an integrated framework based on statistical tests, time-series neural networks, and improved MLP, ANFIS, and SVR with novel meta-heuristic algorithms in order to obtain the best prediction of DPD in Iran. At first, a series of economic and social indicators that seemed to be effective in the demand for dairy products is identified. Then, the ineffective indices are eliminated by using the Pearson correlation coefficient, and statistically significant variables are determined. Since the regression relation is not able to predict this demand properly, the artificial intelligence tools including MLP, ANFIS, and SVR are implemented and improved with the help of novel meta-heuristic algorithms such as grey wolf optimization (GWO), invasive weed optimization (IWO), cultural algorithm (CA), and particle swarm optimization (PSO). The designed hybrid method is used to predict the DPD in Iran by using data from 2013 to 2017. The high accurate results confirm that the proposed hybrid methods have the ability to improve the prediction of the demand for various products.

摘要

这项研究专门针对乳制品需求 (DPD) 的预测。由于乳制品的消费周期很短，因此掌握有关其未来需求的准确信息非常重要。本研究的主要贡献是提供一个基于统计测试、时间序列神经网络和改进的 MLP、ANFIS 和 SVR 的集成框架，并具有新颖的元启发式算法，以获得伊朗 DPD 的最佳预测。首先，确定了一系列似乎对乳制品需求有效的经济和社会指标。然后，利用 Pearson 相关系数剔除无效指标，确定具有统计显着性的变量。由于回归关系无法正确预测这种需求，在新的元启发式算法的帮助下实现和改进包括 MLP、ANFIS 和 SVR 在内的人工智能工具，例如灰狼优化 (GWO)、侵入性杂草优化 (IWO)、文化算法 (CA) 和粒子群优化(PSO)。所设计的混合方法通过使用2013年至2017年的数据对伊朗的DPD进行预测。高精度的结果证实所提出的混合方法具有提高对各种产品需求的预测的能力。