Estimating missing values in China’s official socioeconomic statistics using progressive spatiotemporal Bayesian hierarchical modeling

Abstract

Due to a large number of missing values, both spatially and temporally, China has not published a complete official socioeconomic statistics dataset at the county level, which is the country’s basic scale of official statistics data collection. We developed a procedure to impute the missing values under the Bayesian hierarchical modeling framework. The procedure incorporates two novelties. First, it takes into account spatial autocorrelations and temporal trends for those easier-to-impute variables with small missing percentages. Second, it further uses the first-step complete variables as covariate information to improve the modeling of more-difficult-to-impute variables with large missing percentages. We applied this progressive spatiotemporal (PST) method to China’s official socioeconomic statistics during 2002–2011 and compared it with four other widely used imputation methods, including k-nearest neighbors (kNN), expectation maximum (EM), singular value decomposition (SVD) and random forest (RF). The results show that the PST method outperforms these methods, thus proving the effects of sophisticatedly incorporating the additional spatial and temporal information and progressively utilizing the covariate information. This study has an outcome that allows China to construct a complete socioeconomic dataset and establishes a methodology that can be generally useful for estimating missing values in large spatiotemporal datasets.

Risk Assessment and Mapping of Hand, Foot, and Mouth Disease at the County Level in Mainland China Using Spatiotemporal Zero-Inflated Bayesian Hierarchical Models

“…The monthly climate data in this study was based on the raw data collected from 727 climate stations throughout China from the China Climate Data Sharing Service System [ 16 ]. Data of yearly socioeconomic variables were from the China County Statistical Yearbook, China Statistical Yearbook for Regional Economy, and China City Statistical Yearbook [ 39 ]. We included a total of six climatic variables and fourteen socioeconomic variables as the potential environmental risk factors for HFMD in this study ( Supplementary File S1, Table S1 ).…”

Spatial and Temporal Impacts of Socioeconomic and Environmental Factors on Healthcare Resources: A County-Level Bayesian Local Spatiotemporal Regression Modeling Study of Hospital Beds in Southwest China

“…The most up-to-date satellite remote sensing technology could be applied as a potential solution for tackling such data deficiencies. Another limitation inherent in this study was that the dataset we adopted for analysis failed to provide the most up-to-date information, due to the later revision of China’s county-level statistical yearbooks during which multiple vital variables were removed, and several new variables were added [ 22 ]. Failing to find a solution to this problem, in this study, we had to sacrifice the timeliness of our research in order to keep the space scale to the minimum county level, as well as to ensure the comprehensiveness of socioeconomic factors.…”

“…To be specific, we collected a total number of 32 county-level variables, including 20 socioeconomic factors and 12 environmental factors, as a list of potential covariates influencing healthcare resource allocations in southwest China, which were summarized in Table 1 . Among them, the hospital beds and socioeconomic data were retrieved from China’s first official published county-level socioeconomic statistics dataset, which had been originally collected from the China County Statistical Yearbook, the China Statistical Yearbook for Regional Economy, and the China City Statistical Yearbook [ 22 ]. However, as China’s county-level statistical yearbooks have been upgraded by removing a variety of critical variables after 2003, quite a number of necessary socioeconomic variables cannot be maintained after that time point.…”

“…Lastly, the coefficient of determination ( R 2 ) is commonly used in regression models for evaluating the explained variation or variance, which is defined as the degree to which predicted and actual values are consistent. A higher value of R 2 indicates a greater variation that the model can explain [ 22 ].…”

“…It is noteworthy that all these previous studies failed to conduct comprehensive investigations into hospital beds from a socioeconomic perspective with only very limited socioeconomic factors to be taken into consideration. In an attempt to bridge such research gaps embedded in previous literature, a complete space–time dataset which contains various county-level socioeconomic factors in China was recently developed and proposed by Song et al in a newly published study which is believed to serve as a potent tool for exploring the impacts of various socioeconomic factors on hospital beds in a much more comprehensive manner [ 22 ].…”