인공지능을 활용한 빈집 실태조사 효율화 방안에 관한 연구 : 군산시를 사례로

초록

Due to the recent centralization of the metropolitan area, the number of vacant houses in local cities continues to increase. Accordingly, the government is conducting a survey on vacant houses, but the cost of on-site surveys is high due to the low accuracy of the preliminary survey. Therefore, in order to efficiently conduct an empty house survey, it is necessary to accurately find buildings suspected of empty houses in the preliminary survey stage and take follow-up measures accordingly. This study aims to expand the efficiency of the survey of vacant houses by attempting to estimate vacant houses through artificial intelligence by using data on electricity, water usage, buildings, and socioeconomic variables to help early detection of vacant houses. As a result, Decision Tree Ensemble Models showed the best performance based on Accuracy, F1-score, and AUC scores, and building data in addition to electricity usage were also identified as important variables in estimating empty houses.

키워드