Research On The Multivariate Nonlinear Regression Mode Of Industrial Enterprise Innovation Input And Output In Hubei Province

Innovation ability is the basis of scientific and technological innovation driven development strategy, is the embodiment of national competitiveness, determines the long-term motivation of the national economy and indicates the trend of economic development in the next few years. Therefore, the study of innovation ability is of great significance. For the two regulating body of the government functional departments and enterprises, evaluating innovation ability of the enterprise correctly is the premise of in-depth reform to promote the innovation. In this background, we participated in and completed the project of comprehensive evaluation of industrial enterprise’s innovation ability in Hubei province in2012, and the result is inventoried in the proceedings of the second R&D resources in Hubei province.Input-output model is the important tool in the study of economic and management disciplines, has been widely used in all kinds of comprehensive evaluation and prediction. The index system tends to be more complex in the actual input and output model, and the establishment of the model mostly based on multivariate linear regression method. But the multivariate linear model often has two types of problems, the first one is the choice of model, another is the reliability of data. Such as the different variance problem, variable transform is often needed, then the matched model is no longer a multivariate linear model; and if the data is abnormal value, multivariate linear model will be affected by outliers and produces larger deviation. In this article, we adopt the multivariate nonlinear regression and weighted multiple nonlinear regression method to solve the above problems, and the results are reasonable and the relevant experts also recognize the results.Industrial enterprises in Hubei province are selected in this article as the research object, Data Is derived from Hubei province’s science and technology statistical yearbook from2010to2012, on the basis of constructing a scientific and reasonable evaluation index system of innovation ability, the input-output model of the innovation ability of industrial enterprises in Hubei province is established for research. In this paper, the concrete content is as follows: The first chapter is introduction. In this paper, we discuss the research background, research significance and research status, and describe the main research contents and methods in the project and the research emphasis and characteristics.The second chapter is the introduction of the basic situation of industrial enterprise innovation ability evaluation in Hubei province, including the construction of a comprehensive evaluation index system of innovation ability and basic indicators. The innovation ability of input-output model established in this paper is an important component of the comprehensive evaluation, and the selection of the input and output index has a certain scientific nature and rationality.The third chapter is the multivariate nonlinear regression model theory and its application. This part introduces the general method of nonlinear regression model. Specially the weighted multiple nonlinear regression method is proposed, and the feasibility of this method is illustrated by an example.The fourth chapter is the implementation of Hubei province industrial enterprise innovation input and output of multivariate nonlinear regression model. Firstly, we use multivariate nonlinear regression method to establish input-output model, but for some irrationality indicators of multiple linear regression model, we conducted the corresponding nonlinear processing; Secondly according to abnormal values appeared in residual figure, we remove outliers artificially, and then use the rest of the data to construct input-output model; Finally, we put forward and realized the weighted multivariate nonlinear regression model that using the weighted weight from the definition of residual gap. To do so can avoid the subjectivity and complexity in the process of "removing outliers" in a certain extent, the result is ideal.The fifth chapter is conclusion and prospect. This chapter summarizes the full text conclusion, and discussed the weighted nonlinear regression model in this paper. Empirical analysis on weighted nonlinear regression model has certain feasibility and rationality, but the related theory of properties and better the weighting method is worth further study.