|
KMID : 0379220210410050803
|
|
Journal of Korea Gerontological Society
2021 Volume.41 No. 5 p.803 ~ p.823
|
|
|
Exploration of Factors Affecting Life Satisfaction of Older Adults Using Random Forest
|
|
Suh Ye-Lin
Park Ja-Kyung
Ko Gwi-Young
|
|
|
Abstract
|
|
|
|
The purpose of this study is to identify the variables that affect the life satisfaction of Older Adults by using a random forest, one of the representative techniques of machine learning, and to understand the influence relationship with life satisfaction. For this purpose, the 7th data of the Korean Longitudinal Study of Ageing(KLoSA) was used, and a total of 60 variables including the life satisfaction of older adults were finally used for the analysis. The explanatory variables were largely composed of demographic background, family (children and grandchildren, parents and siblings), health, employment, income and consumption, assets, and subjective expectations. For the analysis method, random forest was used to derive variables with high importance among 59 explanatory variables, and multiple regression analysis was performed to identify the relationship between major variables and life satisfaction. The main results are as follows. First, a random forest analysis using 500 trees and 13 split variables showed that the degree of depression, average monthly allowance, subjective perception of social hierarchy, life expectancy, subjective health, number of grandchildren, age, job expectation, personal wealth, total amount of money received. The relative importance of the 10 variables was the highest. Second, as a result of analyzing the partial dependence on life satisfaction of older adults according to the top 10 variables of importance, the patterns of influence on life satisfaction by each variable were very different. Third, as a result of performing multiple regression analysis by inputting the top 10 variables of importance, it was found that the monthly average pocket money had the greatest influence on life satisfaction, followed by subjective perception of social hierarchy, subjective health, and life expectancy. However, these results may be slightly different from the partial dependence results, indicating that not only regression analysis but also partial dependence analysis should be considered in order to comprehensively examine the effects of related variables on life satisfaction of older adults. Based on the above results, the recommendations of this study are as follows. First, research on the lives of older adults needs to be analyzed from a more comprehensive perspective. Second, in addition to the current working status used in the analysis of this study, additional analysis is needed on the variables related to employment of older adults. Third, based on the research results, policy support for the lives of older adults should be provided from more diverse aspects.
|
|
|
KEYWORD
|
|
|
Machine Learning, Random Forest, Life Satisfaction, Older Adults
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|