Clustering Analysis of the Survey for Mobility Reasons in the US 1999-2017

This paper is to estimate the survey for 98000 addresses from 1999-2017 in United States bureau of Census by using cluster analysis. The analysis is mainly applied by Approximate Covariance Estimation for CLUSTING (ACECLUS), and procedure variables for CLUSTING (VARCLUS) to test some important parameters such as average linkage, two-stage density linkage, Cubic Clustering Criterions (CCC), R-Square, Ward’s minimum variance techniques, as well as Tree procedure for deeper exploring the clusters or variables. After the overall analysis, the results show that there is existence of strong covariate correlation for variables X8 and X15 with respond variable Y (Mobility periods). Hence, Reason “Retired” from survey data is most important impact on mobility other than the reasons “Wanted better neighborhood or less crimes” and “Wanted cheaper housing” that are popular and highly frequent.