【摘要】 进口食品安全风险是一个动态、非线性的过程，单一的模型很难做出准确拟合和预测。以2010-01—2021-08间的进口食品不合格情况数据为研究对象，采用自动回归差分整合滑动平均模型(ARIMA)进行建模，运用机器学习方法中的支持向量机(SVM)算法对模型进行优化，建立ARIMA-SVM组合模型。以平均绝对误差(MAE)、均方根误差(RMSE)、平均绝对百分率误差(MAPE)和判定系数(R~2)等评价指标作为模型的评价指标。结果发现：ARIMA-SVM组合模型比单独运用ARIMA模型和SVM模型建立的模型的精度高，对进口食品不合格情况的短期预测效果更好。更多还原

【Abstract】 The safety risk of imported food is a dynamic and nonlinear process, making it difficult for a single model to provide accurate fitting and predictions.Using data on imported food nonconformities from January 2010 to August 2021,this study models the situation using the Autoregressive Integrated Moving Average(ARIMA) model, optimizing it with the Support Vector Machine(SVM) algorithm from machine learning to establish an ARIMA-SVM combined model.Evaluation metrics such as Mean Absolute Error(MAE),Root Mean Square Error(RMSE),Mean Absolute Percentage Error(MAPE),and Coefficient of Determination(R~2) are used to assess the model.The results show that the ARIMA-SVM combined model has higher accuracy than models established using only the ARIMA or SVM models, providing better short-term predictions for imported food nonconformities.更多还原