所以用这6个关联度系数大的输入因子的(x_h )来择选与预测数据输入因

Therefore, the six input factors with large correlation coefficients (x_h) are used to select historical data that is close to the input factor of the predicted data, thereby reducing the data capacity of the historical data; that is, the |Δ| and μ corresponding to the x_h input factor For comparison, choose historical data with |∆|≤μ. Where μ is the data selection coefficient of the input factor of x_h, which is mapped with |Δ|, and the size of μ needs to be continuously adjusted and optimized; if it is too large, all historical data are easy to pass, and the input factor that is close to the predicted data input cannot be selected. Historical data; μ too small to discard all historical data, resulting in no historical data to find.

Therefore, the historical data of the input factor sicity of these six large correlation coefficients (x_h) is used to select the historical data similar to the input factor of the prediction data, thereby reducing the data capacity of the historical data, and the historical data of the input factor corresponding to the input factor of the x_h is to be compared with the corresponding . Where in the x_h the data selection coefficient of the input factor, and the one-phase mapping, the size of the s needto be constantly adjusted and optimized, the size of the historical data is too large to pass, can not be selected and the prediction data input factor similar historical data, the s is too small to discard all historical data, resulting in no historical data to look for.

So we use the (x_ H) to select the historical data similar to the prediction data input factors, so as to reduce the data capacity of historical data; that is, X_ H input factor corresponding to |||||||||||||||≤ μ historical data. Where μ is x_ The data selection coefficient of H input factor is mapped with | | |, and the size of μ needs to be adjusted and optimized constantly; if μ is too large, all historical data are easy to pass, and historical data close to the input factor of predicted data cannot be selected; if μ is too small, all historical data will be discarded, resulting in no historical data to be found.<br>