Identification Of Poor Households In Precision Poverty Alleviation In Guizhou Province

As the change of current China the whole macro economic environment,especially the level of national income distribution inequality,had extensive regional development as the main pattern of poverty has been showing off target and helping to reduce the effect,and a series of problems.More than 30 years of rapid economic growth made by average level measured along with the increase of the per capita income,high speed,also showing abnormal serious phenomenon of uneven income distribution,and so on.The national Gini coefficient(inequality)increased from 0.29 in 1981 to 0.47 in 2012,the national inequality levels increased by 65%.The Gini coefficient of rural increased from 0.21 in 1978 to 0.39 in 2011,increased by 83%.Inequality levels expanding at the same time,it means that in the poverty population to enjoy economic growth at the bottom of the income distribution of profit will become more difficult,namely through economic growth decline in poverty reduction effect unceasingly.So the future of China's poverty alleviation through economic growth to a regional large-scale anti-poverty mode basically is not feasible.So for more targeted anti-poverty mode becomes more important,precise poverty alleviation in this environment arises at the historic moment.Precision of poverty alleviation is of great significance to can realize a well-off society in an all-round way,but it's also a great complexity of the project,should be accurate there are still many difficulties need to overcome poverty alleviation work.At present the emphasis and difficulty in accurate poverty alleviation work mainly concentrated in the "accurate identification","accurate support",and "precise evaluation".Accurate recognition as the basis of accurate poverty alleviation,can accurate and effective to identify poor is helping precisely,the first premise of accurate appraisal,once poor identification error precision for poverty alleviation and from any talk about poverty.Wang's theory of the precision for poverty alleviation in China "is pointed out that his WuMengShan District in guizhou,yunnan,sichuan province,according to the results of sample survey of the 1200 households of farmers is not documented in 2013 to 58% of the farmers have a tent card,family income per capita is lower than the stipulated by thestate 2300 yuan of the poverty line,and the document in the tent card poor showed that 40% of farmers per capita income of more than 2300 yuan of the poverty line.In wuling mountainous area in guizhou,chongqing,hunan and hubei province in 1000 by inputting tent card poor sampling results showed that only 49% of the farmers income above the poverty line of $2300.Through the survey on income as the judgment standard,democratic appraisal this method leads to the recognition error rate of about 50%.Given by inputting tent card data found that 90.76% of the poor in guizhou is concentrated in the wuling mountain area,dianqiangui rocky desertification area,WuMengShan District the three area.Poor such high recognition error rate keep accurate support and accurate assessment of the subsequent brought a big challenge.Accurate identification of poor,therefore,is a problem to be solved.This article is based on accurate documentation for poverty alleviation in guizhou in 2014 tent card is funding the sampling data,do the work as follows: the first: logistic regression model was constructed.First of all,made detailed analysis of the main characteristics of the poor.Second for logistic poor recognition model in 0.5 as the cutting point and the optimal cutting point to do the contrast on the ability to recognize,the results show that the optimal cutting point of logistic model recognition ability is superior to logistic model with the cutting point of 0.5.Secondly,using the method of random forests to build poor recognition model.On the selection of tree number using fifty percent cross validation method,and draw out the poor and the poor the importance of each variable.Third: poor recognition model was constructed with AdaBoost method.Fourth: comparative study of the four model,according to the results on the funding poor identification capability and stability of the random forest method to build model is superior to the other four models,and significant in statistical sense,AdaBoost and logistic...