Polymers are raw materials for products in our daily life such as plastics, resin, rubber, and organic glasses. The long molecular chains in polymers generate versatile and complex structure in mesoscopic scale, which affects the material property. Therefore, it is hard to deal with the mesostructure-property relationships by simple experiment or theory. Recently, the author and coworkers have developed the mesostructured-property relationships in polymer science based on mathematical and informatics methods. First, description of rubber elasticity of elastomers by complex network analysis are represented. Then, process-mesostructure-property relationships in crystalline polymers by machine learning techniques are mentioned. Furthermore, reverse analysis of biodegradable polymers with both toughness and degradability are included.