Methodological aspects of current statistical survey of population life quality within the unstable regional development conditions

Abstract

The article covers the main approaches to current sample population survey on quality of life problematic in regional and municipal dissection. A block scheme is suggested for the system of indicators of current population survey on the aspects of: demographics and social sphere; environment and healthcare; labour resources; welfare and consumer prices; production and infrastructure. An algorithm of current population sample survey is developed in connection with operative monitoring of regional development core indicators. The statistical instrumentary of multivariate analysis of sample population survey results is suggested, with the listed parameter inputs for the methods applied. For the multivariate analysis the results of hierarchical cluster analysis are presented as to define the population classes within the sampling; at the next stage the k-means clusterization was applied to form the clusters, after which the most significant differentiating indicators are listed with the descriptive statistics for the clusters. There are then defined the opportunities of applying “Random forest” modeling methods for the purposes of classification of the population among the obtained clusters, and the test data were used to construct and probate the “random forest” model for further surveys. To reveal the main factors defining the life quality, a factor analysis was carried out using the maximum likelihood procedure.