In Japan, the ratio of aging sewage pipes is rapidly increasing, and the three shortages of financial resources, human resources, and technology are becoming problems in maintenance and management. Under these circumstances, it is difficult to conduct a comprehensive survey of a massive number of sewage pipes．Therefore, it is necessary to prioritize the inspection and investigation of a massive number of sewage pipes. In determining priorities for sewage pipe that have not been inspected and surveyed, it is effective to estimate the soundness in the sewage pipe. Previous studies have estimated the soundness of pipeline units by using statistical methods and machine learning. On the other hand, inspection plans are often developed on an area level. However, there are no previous studies that have predicted the soundness of sewage pipes on area level. In this study, machine learning is used to estimate the soundness in sewage pipes on a very small mesh area level. The macro soundness estimation method proposed in this study will contribute to the planning of practical inspection plans.