In the site selection of elderly care facilities era, the traditional methods of site selection for elderly care facilities are subjective, the data sources of methods are single, and the research objects of the methods are not realistic, this paper presents a multi-source heterogeneous data-driven method for the site selection of elderly care facilities. This method takes the Xicheng and Dongcheng districts of Beijing as the research objects. On the grid scale of 500m*500m, six features of population factors, economic level, traffic accessibility, medical conditions, the convenience of life, and natural factors are constructed through multi-source data. Using an integrated algorithm, the location of elderly care facilities is predicted quantitatively, and the priority of features is sorted. The research found that: (1) The model in this paper predicts more accurately than traditional models (2) the number of education facilities, proportion of green areas, medical conditions, traffic accessibility, and the number of shopping facilities are the main features for the site selection of elderly care facilities in Xicheng and Dongcheng districts. This method can also provide references for the site selection of other facilities.