基于区间预测模型的流感趋势预测

发布时间：2019-01-01 08:14

【摘要】：在流感暴发趋势预测模型的研究中,传统点预测是估计预测平均值的随机变量,不包含置信水平和预测区间范围等辅助决策的有用信息,导致决策者不能很好把握流感发展趋势。为了解决上述问题,提出利用神经网络上下限估计方法(LUBE)建立预测区间(PI)发展了流感趋势区间预测模型,提出了评价预测区间的宽度范围组合指标(CWC),运用蚁群算法对神经网络区间预测模型进行训练,并运用上述模型对传染病等应急医疗数据进行了仿真。为了衡量预测区间性能,改进模型与Delta、Bayesian、Holt指数平滑和支持向量机等常用预测模型建立的预测区间进行了对比。结果表明蚁群算法神经网络区间预测模型能够对流感趋势进行更为有效的分析和预测。
[Abstract]:In the research of influenza outbreak trend prediction model, the traditional point forecast is a random variable to estimate the average value of the forecast, and it does not contain useful information for auxiliary decision, such as confidence level and range of prediction interval, etc. As a result, policy makers can not grasp the trend of influenza. In order to solve the above problems, a prediction interval (PI) model is developed by using the upper and lower bound estimation method of neural network (LUBE) to develop the interval forecasting model of influenza trend. The combined index (CWC), of the width range for evaluating the prediction interval is proposed. Ant colony algorithm is used to train the interval prediction model of neural network, and the above model is used to simulate the emergency medical data such as infectious diseases. In order to evaluate the performance of prediction interval, the improved model is compared with the prediction interval established by Delta,Bayesian,Holt exponential smoothing and support vector machine. The results show that the interval prediction model of ant colony algorithm (ACA) neural network can effectively analyze and predict the trend of influenza.
【作者单位】：上海交通大学机械动力工程学院;
【分类号】：R181.3;TP18

【参考文献】