Research And Application Of Lasso Regression Model Based On Prior Sparse Frame

With the rise of the era of big data in recent years,the analysis and processing of data has received increasing attention in various scientific fields such as social sciences,information science,genetics,biology,medicine and finance,so how to deal with massive data The extraction of its essential features has become an important research direction.And in terms of massive data,how to build a suitable data model and mine data with a small number of features but relatively comprehensive information for analysis is a problem that every data scientist needs to face.Among the many data models,the Lasso model is a model that can effectively process high latitude data without losing the corresponding accuracy.It is a typical variable selection method,which can compress some small variables to 0 and eliminate redundant variables by setting the threshold and limiting the size of the parameter sum.For the traditional regression model,the Lasso regression model and its improved model can solve the problem of variable selection well.Therefore,the Lasso method and its improved method have received great attention in statistical research.In this paper,a new Lasso improvement method is proposed for the Lasso regression model.The prior information is incorporated into the Lasso regression model.This paper refers to the Lasso regression model based on the prior sparse framework.First of all,this paper introduces the advantages of the Lasso regression model on the regression problem compared with other models,and explains the algorithm of the Lasso regression model and its good properties.Secondly,this paper introduces the concept of sparse framework and Lasso regression model,and introduces a more general Lasso regression framework.It can be converted into a variety of known Lasso variants.Again,the prior information is part of the free attribute of the regression analysis feature,which itself cannot be used to describe the research object,so it is necessary to apply the prior information of the first feature itself to the model in a unique way.Therefore,this paper cites the general Lasso regression framework,and gives the definition of Lasso regression model based on prior sparse framework,and theoretically gives the corresponding algorithm and properties.Finally,this paper analyzes multiple sets of simulation data and empirical data.The results show that the Lasso regression model based on prior sparse framework has better performance than the ordinary Lasso regression model with prior information.