Quasi-likelihood Estimation In Generalized Semi-functional Partial Iinear Regression

With the development of society, the computer storage capacity and enhance the processing speed, we observed an increasing number of areas of environmental science, chemistry, biology, medicine, economics and other data more and more sophisticated. For example, a phenomenon we can observe a large sample of variables, we further look at a common scenario:A random variable can take the values at some point of time range, it’s an observation sample can be represented by a random family. In modern statistics, observational data for a given range means that more and more continuous constant closer and closer. Traditional statistical methods and statistical models, there are many problems in dealing with such data, such as over-fitting and dimensionality curse problem. In order to solve these difficulties, Statistics scholars to consider these observations of a large sample of data into a continuous family of each individual as a curve, and thus the curve data for statistical analysis. This is the basic idea of this article function type data.Partially linear model theory first by Engle et al (1986) suggested that followed has been widely studied and applied in many areas of applied statistics. This model allows parameters form part of the explanatory variables, while the other part is non-participation in the form of explanatory variables. Thomas then put this model to the generalized formIn this paper, quasi-likelihood approach to estimation of the generalized partially linear model and the recent development of function-type data in nonparametric statistics, the data type of the function is introduced to estimate the explanatory variables in the past, part of Generalized Linear Model Semi function for some progressive nature of the model parameter estimators have been described. Finally, a real example to illustrate the value of the estimated effect of the model in this paper.