Hypothesis Test Of Zero-and-one Inflated Poisson Regression Model

In statistical research and application,linear regression model is an important parametric regression model,which is the basis for studying other statistical regres-sion models.Linear regression models are also widely used,such as in economics?finance,medicine?industry?agriculture and other fields.However,if the data does not obey the linear regression model,it still uses it to statistically infer the data,perhaps not revealing the true correlation hidden in the data,causing a mis-understanding of the thing judged,resulting in the prediction of the things being judged lost accurately.Sex.Therefore,it is very important to study whether the data is subject to linear regression models or a given parametric regression model structure:Y=f(X)??:,If the form of the regression function f(X)is known,only the unknown parameters,the parametric regression model is used to study the da-ta.If the form of the regression function is completely unknown,the nonparametric regression model is used.Analysis,and then explore the structure of the model.In the practical application and statistical analysis,the function form of the regression model is generally unknown,so the nonparametric regression model is commonly used for research.There are many methods for estimating the regres-sion function using the nonparametric regression model.The commonly used kernel method and local ploynomial method are used,spline methods,etc.The main re-search problem in this paper is to assume that a set of data has been estimated based on the local ploynomial method(the main discussion is the local linear re-gression method),and the results of the parameter estimation by local regression are analyzed,for example,The scatter plots of the values show that these fitted curves may be a straight line,and thus the research problem of this paper is proposed:if the observed data obeys the linear model,then the theoretical expectation of local parameter estimation is the same,and the estimated values are not much different.Thus the linear problem of the test observation data is transformed into the differ-ence problem of the test local parameter estimation.In this paper,the appropriate test statistic is constructed for the proposed test problem,and its properties are disscussed.The theoretical results are verified by data simulation,which shows that the method is feasible.The innovation of the thesis is that the regression test is not based on the orig-inal observation data construction test statistic,but the local parameter estimation is obtained by local linear regression,and the hypothesis test is constructed based on the local parameter estimation construction test statistic.Obviously,if we have original observation data,we can directly construct a global linear regression model to complete the linear test of the model,and the test statistic proposed in this pa-per is constructed based on local linear regression estimation.Obviously it seems that there is no good test statistic based on the original observation data.But the significance of this is that even if the observed data is lost only under the condition of its local parameter estimation,the data can be linearly tested.In the context of the rapid development of modern information technology,the amount of raw data may be very large,difficult to store or there is a lack of,the problem of research in the text has a certain significance.