LAUSR

Abstract A comprehensive database containing data on approximately 2700 buildings and 6000 full-scale measured period samples was constructed through massive literature searching and stringent data filtering. The newly emerged maximal information coefficient method, which is suitable for large data set statistical analysis, was adopted in conjunction with Kruskal–Wallis analysis of variance to identify factors that significantly affect a building’s fundamental period. It was quantitatively verified that height, predominant structural material, and lateral-force resisting system are the three most important influencing factors. Subsequently, height was used as the dominant regression variable, and material and lateral-force resisting system were used as categorical variables, predictive models in combination with confidence intervals of the fundamental period are provided for multi-factors, including four material types and three structural types. In addition, multi-level empirical formulas of the natural period in other five modes (two translational and three torsional) are provided on the basis of the regression results of the fundamental period. All these predictive models can effectively reflect the tendency of the median and the rational scope of variability of the natural period of buildings.