ScholarWorks@Y-Scholar Hub

초록

This study aims to explore the extinction risk of local cities and counties in Korea. For analysis, it uses factors that affect the risk of extinction or the attractiveness of the region. and data is collected and organized in the KOSIS, and then, the improved extinction risk index is derived for local regions by reflecting population movement. the extinction risk index is utilized by dependent variables (target value). and we construct a machine learning analysis model with the independent variables(features) and dependent variables. Machine learning models are GBM, RF, XGB, etc. and they predict and classify local decimation risk in a way that improves the performance of the model with enhancing methods like voting and ensemble. The results derive the local region with the highest risk of extinction and determine the factors that affect the local extinction risk. As a result of the analysis, the prediction performance of the machine learning model built in this study was around 90% and 68 local regions were measured with the highest risk of extinction. Factors affecting these extinction risks were found to affect economic and industrial factors, and convenience factors such as cultural and medical facilities. In conclusion, to overcome local extinction in the future, local governments in danger of extinction need to focus first on these economic and industrial factors improved, and if those factors are overcome, it is important to expand cultural and medical facilities to enhance local attractiveness.