| 凌立文 1,2，徐镁淇 1，张学竞 1.基于多维关联规则的猪肉价格波动影响因素分析及预测建模研究[J].广东农业科学,2022,49(12):167-175 |
|
查看全文
HTML
基于多维关联规则的猪肉价格波动影响因素分析及预测建模研究 |
| Analysis of Influencing Factors of Pork Price Fluctuation and Prediction Modeling Based on Multidimensional Association Rules |
| |
| DOI：10.16768/j.issn.1004-874X.2022.12.019 |
| 中文关键词: 猪肉价格 影响因素 多维关联规则 支持向量回归机 价格预测 生猪疫病 |
| 英文关键词: pork price influencing factor multi-dimensional association rule support vector regression price forecasting pig epidemic |
| 基金项目:国家自然科学基金（71971089，72001083）；广东省自然科学基金（2022A1515011612） |
|
| 摘要点击次数: 763 |
| 全文下载次数: 521 |
| 中文摘要: |
| 【目的】我国是猪肉生产及消费的大国，近年来，猪肉价格波动呈现频率加快、幅度增大的趋势。猪肉价格波动不仅增加农户收益的风险性，也在一定程度上影响广大民众的生活。正确识别猪肉价格波动的影响因素并对猪肉价格波动进行科学预测，有助于确保市场健康平稳运行。【方法】运用多维关联规则定量分析生猪养殖加工产业链、替代品市场、宏观经济环境变化、突发性事件和国际市场环境等 5 方面共 16 种因素与猪肉价格波动的关联和影响程度，将挖掘得到的高相关因素作为模型输入变量，运用支持向量回归机构造提前多步的猪肉价格波动预测模型。【结果】与猪肉价格波动关联程度最高的前 3 位因素是生猪疫病、生猪价格和仔猪价格，置信度分别为 1.00、0.93 和 0.82；对猪肉价格影响程度最大的前 3 位因素是生猪疫病、猪肉产量和出栏猪肉量，提升度分别为 1.84、1.67 和 1.67。相较于基准预测模型，将 12 个高相关影响因素作为模型输入，均方根误差减少 29.11%，平均绝对百分比误差减少 16.00%。【结论】使用多维关联规则进行变量筛选，不仅能减少模型的变量个数，还能有效提高模型的预测精度。鉴于生猪疫病对猪肉价格波动的关键影响作用，政府相关管理部门应提高对动物疫病的风险防范意识。 |
| 英文摘要: |
| 【Objective】China is the main country in pork production and consumption. In recent years, pork price fluctuations have shown the trend of accelerating frequency and increasing amplitude. The fluctuation of pork price not only increases the risk of farmers’ income, but also affects the living of the general public. Correctly identifying the influencing factors of pork price fluctuations and making accurate predictions help to guarantee the well-functioning of the market.【Method】The multi-dimensional association rules were used to quantitatively analyze the correlation and influencing degree of 16 influencing factors in five main aspects including the industrial chain of pig breeding and processing, market of substitutes, changes in macroeconomic environment, emergencies and international market environment with pork price fluctuations. High correlation factors selected by mining were used as model input variables, and the support vector regression
machine was used to construct a multi-step prediction model for pork price fluctuations.【Result】The top three factors most related to pork price fluctuations are pig epidemics, pig prices and piglet prices, with confidence levels of 1.00, 0.93 and 0.82 respectively. The top three factors have the greatest impact on pork prices are pig epidemics, pork production and pork supplied to the market, with the improvement degrees of 1.84, 1.67 and 1.67, respectively. Compared with the baseline prediction model, with 12 highly correlated influencing factors as model inputs, the Root Mean Square Error (RMSE) is reduced by 29.11%, and the Mean Absolute Percentage Error (MAPE) is reduced by 16.00%.【Conclusion】The use of multi-dimensional association rules for variable selection not only reduces the number of variables, but also effectively improves the prediction accuracy. Given the vital influence that pig epidemic imposes on price volatility, authorities should raise the awareness of risk prevention of animal diseases. |
|
查看/发表评论 下载PDF阅读器 |
|
|
|
