关键词: 通勤距离, 决策树, 决策规则, 北京

Abstract:

With the development of suburbanization, urban residents now have more choices in jobs and housing locations. Nowadays, scholars increasingly pay attention to the studies on citizens' commuting mode. The analysis of commuting space characteristics belongs to the study of geography. Based on questionnaire survey, this paper first makes a descriptive analysis of people's commuting variables, distances, and directions. Then it discusses the commuters of Beijing by decision tree analysis and data mining. Conclusions are obtained as follows: First, under the fixed pruning severity, people's commuting distance is related to their traveling vehicles, resident locations, jobs, youngest child's education conditions, living space, family incomes, usage of cars, and employment rate on local areas. Factors such as gender, educational level, marital status, housing property are not involved in the mode. Second, our study of the relations between the eight variables and commuting distance is different from some foreign studies. Some of these variables are decisive to long-distance commuters while others are decisive to short-distance commuters. Since the housing property has various types such as commercial housing, affordable housing, leasing of public housing and so on, the housing space does not have obvious effect on long-distance or short-distance commuters. Among all the reasons, migrating is not the main cause of shortening the commuting distance. Those who do not migrate have longer commuting distance because they work in suburban areas. The family life cycle with children in kindergartens or primary schools is not a key factor in shortening the commuting distance as the situation in western countries, because there are boarding nursery kindergartens, welfare on giving birth and nursing, which is different from China. Also, wage earners and specialists are not the key factors effecting commuting distance, because of the passive suburbanization in Beijing.

Key words: commuting distance, decision tree, decision rules, Beijing