Study On The Measurement Of Population Mobility Based On Big Data Of Mobile Communication

Population mobility is an important indicator of economic and social development,used to measure the flow of people to pursue economic and social goals and thus form a long period of free movement and off-site life situation.In accordance with the government statistics,the fooating population refers to the population under the conditions of the Chinese household registration system,leaving the residence to other places to live,but there is no clear,accurate and unified definition.By the end of 2016,the total number of floating population in China was about 245 million.Economic growth is an important reason for the flow of population.In view of the complexity of the composition of the floating population,the uncertainty of the flow cycle and the change of the trajectory of the flow,there are many problems in the existing population mobility statistics in China,the caliber of different statistics,uneven data quality,cannot meet the statistics needs of the government and society.The statistical methods and related systems of floating are needed to be improved.Based on the real-time call record data of mobile communication operators and the characteristics of population behavior,the paper makes a judgment and measures the mobility of the population from the user behavior represented by big data of mobile communication.Based on the further definition of the concept of floating population,the paper designs a method to estimates the size of floating population by combining the mobile population identification model based on the machine learning algorithm with the population liquidity measurement model based on the capture re-capture sampling method.In the process of constructing the mobile population recognition model based on the machine learning method,the paper constructs the floating population identification characteristic variables by analyzing behavior characteristic of the floating population and the local population of the mobile communication user and the AUC-RF is used to select the characteristic variables.On this basis,we choose Decision tree,Bagging,Random Forest,Support Vector Machine and artificial neural network algorithm to build the model.And through a variety of evaluation criteria to evaluate and select the model,we finally choose Random Forest model with the best classification performance and generalization ability as the final mobile population identification model and then to achieve the sample set of unclassified samples of classification prediction.In the construction of the population size measurement model based on the capture re-capture sampling,the empirical results show that the estimation method can estimate the size of the floating population accurately and reliably.Therefore,the paper concludes that the method of population mobility measurement based on mobile communication data can be used in parallel with the traditional floating population survey methods,and these two methods complement each other and confirm each other.The paper expects to explore a statistical method and system based on Big data thinking to improve the investigation of floating population in China on the basis of the big data of mobile communication.It relies on the same period of mobile communication record data and uses the scientific statistical inference method to estimate and extrapolate the size and characteristics of the floating population,resulting in more accurate and complete demographic data.The empirical test shows that the method is of low cost,fast speed and high precision,which is very suitable for the improvement and expansion of the current statistical system in China.