Combined Prediction Model Based On Generalized Function Integration Operator And Its Application

The prediction model includes single prediction and combination prediction.Combinatorial prediction refers to a combination forecasting model based on the different characteristics of things which reflects in different single predictive models and weighted average.The results of a single predictive model may be good or bad.Therefore,combinatorial prediction,as an important subject of predictive research,has been widely used in many fields of real life.There are two key points in combinatorial prediction: one is the combination method(how to combine)and another is the determination of the combination weight coefficient(how to get the weight).The way to combine is mainly to study the problem of "How to combine" and "Why can it be combined like this"(the rationality of the combination).The calculation of weight coefficient is the problem of how to design the minimum criterion of error(or the highest precision).In this paper,the two aspects are studied separately,and a new combination method and a new weighting coefficient determination criterion are put forward.Mainly include:(1)Combining tangent function with integration operator,this paper puts forward "tangent type integration operator",explores their special properties of this kind of integration operator,and establishes the combined forecasting model based on tangent integration operator,and carries out the empirical analysis.(2)In the process of studying "tangent integration operator",it is also found that the general monotone function can also be combined with the integration operator,and then the exponential integration operator,logarithmic integration operator and tangent integration operator are generalized to the general "function-type integration operator",and a generalized weighted ordered function average(IGOWFA)operator is proposed.According to the different methods of "function",the existing real-type integration operators are classified.The properties of IGOWFA operators are studied,and the combined forecasting model is established based on the IGOWFA operator,and the example is tested.(3)On the basis of second-order predictive validity theory,the concept of "predictive variability" and "prediction Error degree" are proposed.At the same time,we put forward two new criteria of combinatorial forecasting model: Minimum criterionof predicting variability and minimum degree of prediction error.And based on these two new criterias,we set up a combined forecasting model,discuss the properties of these two kinds of combination forecasting models,and carry on the empirical analysis respectively.