|
KMID : 1124020190350030101
|
|
Korean Social Security Studies
2019 Volume.35 No. 3 p.101 ~ p.127
|
|
|
The Determinants of Subjective Class Identification in Korea: Exploratory Data Analysis
|
|
Kim Won-Gyoung
Choi Pil-Sun
|
|
|
Abstract
|
|
|
|
In this exploratory study, we analyze factors that affect individuals’ subjective class identification in Korea. While the majority of previous studies are based on a specific theory or a priori judgment, our approach is different: we begin with utilizing all available variables. First, we have employed the exploratory data analysis (EDA) to find out all possible variables that affect individuals’ subjective class identification. And then, a specific relationship is identified through the regression analysis. Using the 2017 Social Survey data provided by Statistics Korea, we extract the main explanatory variables from a total of about 70 variables using a random forest method of machine learning. The regression analysis is then used to examine the sign and statistical significance of the coefficients. The results suggest that individuals’ subjective class identification is influenced not only by certain factors, such as income or consumption but also by other various aspects existing in the social context.
|
|
|
KEYWORD
|
|
|
subjective class identification, exploratory data analysis, random forest, ordered logit model
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|