A System for Determining Road Prices Based on Machine Learning Using Public and Private Sector Data

Fixed asset tax is a very important source of funds for Japanese municipalities. In the works of local governments, road price is one of the indicators that determine fixed asset tax. Every three years, the government inspects each road to set an evaluation value (per square meter), which contribute to the fixed asset tax for residential lands facing the street. Municipal staffs spend tremendous time and effort on field surveys to conduct appropriate evaluations. Thus, it is necessary to develop an efficient evaluation system utilizing digital technology. This study proposes a system that automatically estimates the road price. It utilizes the gradient boosting decision tree, which is one of the machine learning methods, to analyze public and private sector data possessed by local governments. In addition, we adopt the judgment basis visualization method to reduce the burden of road price formulation work and improve work efficiency and transparency. Through experiments, it has been confirmed that this system effectively increases work efficiency for the generation of road prices.