Relation Analysis Based On BP Net And It's Application In Economics And Management

Artificial neural network is a complex network system, artificial neural network especially BP network has the function of parallel handling, fault tolerance, self-learning, etc. So it is more feasible for modeling for economic and management problem than econometrics, but neural network model is a "black box", training date that obtained through practice can train network to make the system studied accurate, and get the system input and output relation, but the model result couldn't be explained. In the sturdy of a concerned economics or management system, we not only want to set up the relation between input and output, but also expect to reheal the inner function through this model. So this thesis puts forwards a relation analysis method which is based on BP network, which is to resolute the partial derivative of model output variable versus input variable using chain rule, the partial derivative reflects the relationship of model output variable and input variable, thus provide a reasonable economic explanation. This article also elaborated the application of the connection analysis in the economics and management.