欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
变权重组合预测相比单预测、定权重组合预测能提高预测精度，其关键是确定权重。变权重组合预测模型权重的确定有很多方法，而预测期权重的确定很少受人关注。已有的确定预测期权重的方法是分别对每种预测方法的权重作预测，再归一化处理。文章基于成分数据的知识来确定预测期的权重，与已有方法相比，该方法不仅能保证预测期的权重求和为1，而且能降低维数。实验分析表明：分别模拟50，100，150，…，1000次该方法预测精度比已有方法高，所占的比例大于50％；实例分析中该方法的MSE小于已有方法的MSE。从实验分析和实例分析可以看出，该方法能提高预测精度，是一种行之有效的方法。
英文摘要：
Compared with the single prediction, combination forecast with fixed weight, the combination forecast with variable weight can improve prediction accuracy, which focus on the determination of weight. There are many methods for the weight of model, but few people pay attention to the predicted weight. The previous method of determining the predicted weight is respectively forecasting the weight of every prediction method,then normalizing the weight. This paper proposes a new method that determining the weight based on the compositional data, compared with the previous method, this method can not only ensure that the sum of weight is 1, but also can reduce the dimension. Experimental analysis indicates that respectively simulate 50,100,150,…, 1 000 the proportion that this method has higher prediction accuracy than the previous method is greater than 50%. In example analysis,the MSE of this this method is less than the MSE of previous method. Experimental analysis and example analysis show that this method can improve prediction accuracy, which is an effective method.
同期刊论文项目
同项目期刊论文
An initialization method to simultaneously find initial cluster centers and the number of clusters f
期刊信息
位置：