Data-driven historical preservation: a case study in Shanghai

Historical preservation is becoming ever important in globalizing Shanghai city. However, traditional survey-based ways of policy making are not efficient. This work introduces the data-driven technique with machine learning algorithm to find the relationship between the features of the historical sites and the popularity, which relates to the economy such as tourism and the associated GDP contribution. The method is automatic, which relieves the work load from statistical surveys and other inefficient traditional approaches. Moreover, while the surveys can only reflect the current conditions, the machine learning approach has the ability of predicting the possible outcomes based on existing data, which is helpful when decisions on protection and development are to be made. We collect data from selected historical sites in Shanghai to illustrate the procedure of the proposed data-driven approach. The case study demonstrates the capability of prediction and shows its promising future in guiding policy making, resource allocation and scientific research.