As the population in urban areas increases, the importance of adequate public-led development policies for sustainable cities with affordable housing is becoming more highlighted. In this regard, this study aims to determine the effectiveness of public-led urban development policies for sustainable growth in urban areas, specifically measuring the effect of housing site development projects on housing prices. The geographical scope of the study is the project sites and their surrounding areas in South Korea, and the temporal background is from 2006 to 2023. The project sites were subdivided into four groups by using the Self-Organization Map (SOM), a machine-learning-based clustering analysis, to collect characteristics of each region. Then, the impact of the policy and the prediction of the real estate market of each cluster were analyzed by applying the DID and LSTM models, which have recently been proven to show a high validity. The results show that each cluster had different characteristics and effects from the development projects, depending not simply on the location, but on several characteristics, including the level of size, infrastructure installation, input cost, etc. Furthermore, it is expected for future studies that more detailed research should be conducted with larger datasets of the regional characteristics.