Logistic regression-based automatic lithological naming method

The invention discloses a logistic regression-based automatic lithological naming method. The method comprises the following steps that: 1) the mineral component percentage data of the rock sample of a to-be-named stratum are obtained, the structural characteristics of granularity, rounding and cement components are observed by naked eyes; 2) rock sample data with thin section authentication and naming conclusions are collected as a training set; 3) with the to-be-named stratum adopted as a test set, a thin section authentication conclusion as a dependent variable, granularity, rounding and cement components as factors, and the mineral component percentage content as a concomitant variable, unordered multi-classification logic regression is carried out, logic regression models of various lithological properties are established; 4) the mineral component percentage, granularity, rounding and cement component data of the to-be-named stratum are substituted into logic regression models; and 5) lithological naming