Abstract: How to reduce the dimension of multi-dimensional data becomes more and more important in the era of big data. The traditional method of data dimensionality reduction can not understand all the dynamic trends at the same time, therefore has certain limitations. This paper adopts the dynamic factor method designed for panel data, and take the measurement of urbanization quality of Yangtze River Delta urban agglomeration as the case to carry out the dimensionality reduction experiment of multi-dimensional social data It is found that this method can more accurately identify the “latent variables” which can affect multiple variables at the same time in multi-dimensional data, and can describe the variation trend of  “latent variable” of each individual sample along with time and that of the whole sample. Combined with the results of case study, this research found that the Yangtze River Delta urban agglomeration also exists the problem of incongruous between population urbanization and land urbanization, which needs to accelerate the construction of New-type urbanization and focus on improving the quality of urbanization.