%0 Journal Article %T 基于ELM-GBDT组合模型的生鲜农产品线上需求趋势预测研究
Forecasting Online Demand Trends for Fresh Agricultural Products Using an ELM-GBDT Ensemble Model %A 盛依勤 %J Modern Management %P 38-51 %@ 2160-732X %D 2025 %I Hans Publishing %R 10.12677/mm.2025.155126 %X 在生鲜农产品销售渠道中，由于生鲜农产品的特殊性，需要对于市场信息有较强的感知能力以协调供应链，促进生鲜农产品供应链的高效稳定运行。而供应链特别是上游在市场实时信息对接方面存在明显不足问题。为此，本文借助面向全平台公开的评论数据，以提高供应链信息传递的实时性与准确性，利用ERNIE模型进行评论文本的情感分析，获取用户情感特征集，并依据现有数据进行特征工程，保留提取6个特征后，通过PSO优化获取权重组合ELM-GBDT预测模型进行需求趋势预测。研究结果表明，与单一模型相比，组合预测模型显著提高了对生鲜农产品销量需求趋势的预测准确度，同时发现生鲜农产品需求趋势变化存在一个相对稳定的周期，为生鲜农产品供应链采购决策、库存优化决策等提供有价值的参考。
In the sales channels of fresh agricultural products, due to the unique nature of these products, a strong market information perception capability is required to coordinate the supply chain and promote its efficient and stable operation. However, significant deficiencies exist in the real-time integration of market information, particularly in the upstream segments of the supply chain. To address this issue, publicly available review data across all platforms was leveraged in this paper to enhance the real-time accuracy of information transmission within the supply chain. The ERNIE model was used for sentiment analysis of review texts. A set of user sentiment features was extracted, and feature engineering was conducted based on existing data. Six key features were retained. Subsequently, through Particle Swarm Optimization (PSO), a combination of weights was obtained to construct an ELM-GBDT ensemble prediction model for demand trend forecasting. The research results indicate that, compared to single models, the ensemble prediction model significantly improves the accuracy of forecasting demand trends for fresh agricultural products. Additionally, it is found that there is a relatively stable cycle in the demand trends for fresh agricultural products, which provides valuable insights for procurement decisions and inventory optimization in the supply chain. %K ERNIE情感分析， %K 特征工程， %K 机器学习， %K 生鲜产品， %K 需求预测
ERNIE Sentiment Analysis %K Feature Engineering %K Machine Learning %K Fresh Products %K Demand Forecasting %U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=113867