| China’s big data industry is developing strongly and has become a key industry to boost economic and social transformation.At the same time,the rapid expansion of e-commerce market and the high popularity of electronic payment have also enabled consumers to gradually change their consumption patterns,and the e-commerce platform has thus accumulated a huge amount of transaction data.In view of this,this thesis introduces big data from e-commerce platforms to compile smartphone price indices,with a view to achieving the goal of efficiently serving economic statistics with big data.The smartphone,the object of this thesis,is a typical heterogeneous commodity.To make the smartphone price index more accurate and valid,it is necessary to ensure that the samples are homogeneous and comparable.The characteristic price method is a more effective quality adjustment method,which can well strip out the impact of price changes caused by changes in commodity quality on the price index.Therefore,it is reasonable to combine the big data of e-commerce platform and characteristic price theory into the compilation of smartphone price index.Firstly,this thesis analyzes and summarizes the characteristic price theory and the big data theory in detail.In terms of characteristic price theory,the basic theoretical framework constructed by classical consumer behavior theory,Lancaster’s consumer preference theory,Rosen’s supply-demand equilibrium theory and Diewert’s characteristic price theory is explained,and four manifestations of characteristic functions are compared and contrasted.Then,we compare and review different methods of compiling characteristic price indices from the perspectives of implementation principles,advantages and disadvantages,and applicability,laying the foundation for the later empirical analysis.In terms of big data theory and technology,a preliminary overview of big data is given from the perspectives of definition and characteristics,and then the web crawler technology used in this thesis is developed and discussed,and finally the challenges and innovations brought to economic statistics in the era of big data are discussed.Secondly,this thesis used a web crawler to collect the attribute information and transaction data of smartphones from Tmall e-commerce platform from February 2021 to September 2021.In order to avoid redundant and invalid data affecting the accuracy and validity of the price index compilation,this thesis first preprocessed the crawled big data and finally obtained19,578 valid transaction data,covering 127 smartphone models on sale under 7 brands,and then made detailed observations on the data distribution and characteristics to initially speculate The data distribution and characteristics were then observed in detail,and the factors that may have an impact on the price of smartphones were initially estimated.Finally,dummy variables were introduced to quantify all the feature factors,and then the initial feature screening was performed using the Lasso method first,and then the results of the Lasso method combined with the XGBoost algorithm were used to evaluate the importance of the feature factors,and the feature variables that entered the feature price model were selected comprehensively.Comparing the statistical indicators of the model results under different functional forms,a log-linear function was selected.Then,we construct the characteristic dummy time variable model and the characteristic price model based on the direct characteristic method,and compile the corresponding characteristic price indices,and then analyze the model results while comparing the price indices compiled by the characteristic price method and the simple arithmetic average price index.Based on the above research,this thesis draws the following conclusions: firstly,the impact of each characteristic variable on smartphone prices is different;secondly,the effect of each characteristic variable on smartphone prices varies over time;finally,the performance of the combined characteristic price index is better,which can more truly reflect the changes of smartphone prices on Tmall e-commerce platform during the period and has advantages in robustness. |
