| Over the past few years,anti-poverty has been a global task and a difficult problem.Apart from the great attention paid by the government,how to solve the problem of poverty is also an important issue that the academic circles around the world are devoting themselves to.Under the new situation,in order to improve the efficiency of poverty alleviation work,China has put forward a precise poverty alleviation work mechanism based on accurate identification of poor households and core.Accurate poverty alleviation has become China's top priority and the first livelihood project during the 13 th Five-Year Plan period.However,at present,in the identification of poor households,China still relies on a single income dimension as the basis of evaluation,that is,it relies on the annual per capita comprehensive income index to identify poor households.In fact,poverty is the result of multi-dimensional factors.The one-dimensional identification method is more difficult to comprehensively reflect the poverty situation of farmers,and it is even more difficult to fully identify the causes of poverty.Therefore,in order to more accurately identify poor households,accurately diagnose the causes of poverty,accurately formulate poverty alleviation programs,improve the effectiveness of poverty alleviation work,and truly achieve “real poverty alleviation and poverty alleviation”,this paper is based on the basic information of “three rural” in a city in Yunnan Province.The 2016 data collected in the system construction project is the original sample data,and the purpose of multi-dimensional accurate identification of poor households to help the precise poverty alleviation work,select the year in the six dimensions of income,education,housing construction,agriculture,employment and family size.Nine indicators of per capita income,number of families in school,area of house,whether it is dangerous,area of contracted land,area of contracted forest land,number of family laborers,number of households and actual number of families,firstly,we make full use of data mining and analysis technology,adopt artificial neural network model,support vector machine model and random forest model,and use R language to realize the simulation and accuracy of the three models,in order to analyze the importance of each poverty index;Secondly,establish a multi-dimensional evaluation index system for farmers' poverty,use the two-layer weight distribution method combining entropy weight method and prior knowledge to determine the final weight of each indicator,and construct a multi-dimensional accurate identification model of farmers' poverty,modeled by comprehensive evaluation method.The principle method is used to implement the algorithm.The two-step clustering and K-means clustering analysis method are used to determine the classification of the comprehensive evaluation value,and use this as the basis for evaluation;Finally,in the RStudio development environment,the Shiny software package of the development web interface in R language is used to realize the multi-dimensional accurate identification of poor households service platform,and carry out empirical research on multi-dimensional identification of poor households.The main research conclusions are as follows:(1)Through the comparative analysis of the simulation accuracy of the three models,the results show that the random forest model has the lowest error rate,and then the model is used to derive the importance degree of the indicator.The empirical analysis of the multi-dimensional identification of the pre-treated sample data shows the contribution rate of the various dimensions of the multi-dimensional poverty of the farmers.On the one hand,the scientific evaluation criteria are established to accurately judge whether the farmers are poor,and on the other hand,the precision is selected.The identification method is to achieve accurate identification of poor households.(2)The empirical results of multi-dimensional accurate identification of household poverty show that from the perspective of multi-dimensional poverty-causing factors,the data analysis of effective samples shows that when the number of indicators is 3 and 4,the contribution rate of indicators in multi-dimensional poverty is 32.34% and 24.65%,while when the number of indicators is 5,6,7,8 and 9,the contribution rate of poverty-causing households is 8.39%,2.50%,1.65%,0.033% and 0,respectively.With the increase of the number of indicators in multi-dimensional poverty,the index contribution rate of farmers' poverty has gradually approached zero,which indicates that the effective sample farmers in the selected study area rarely occur in 6,7,8 and 9 indicators simultaneously leading to the phenomenon of poverty.According to the contribution rate of the single consistent poverty indicator,the average contribution rate of farmers' poverty is the highest in the contracted forest land index and the contracted land index,which are 68.50% and 54.64% respectively,followed by the number of households and the annual per capita income indicator for the farmers.The average contribution rate of multidimensional poverty reached 45.72% and 45.39%.(3)The accuracy rate of the multi-dimensional accurate identification model for poverty degree of rural households is 90.26%.Through the analysis of the accuracy and multi-dimensional indicators of model identification and the contribution rate of farmers' poverty,the poverty level of farmers under various dimensions is obtained,so as to accurately identify poor households and their poverty.The results show that the multi-dimensional accurate identification model of rural households has accurate recognition ability and application value in the identification of poor households.Through the realization of the model algorithm,a good poverty is created to accurately identify the application environment,to help the poor work and improve the accurate identification.The accuracy rate of poor households provides technical support. |
PDF Full Text Request