岩体分级的多分类有序因变量Logistic回归模型

将多分类有序因变量的Logistic回归分析引入到岩体质量分级问题中，以影响岩体级别的单轴抗压强度、岩体声波纵波速度、体积节理数、节理面粗糙度系数、节理面风化变异系数和透水性系数为自变量，岩体级别为响应变量，以工程实测岩体质量数据作为统计样本，建立了岩体分级公式。对模型进行了拟合优度检验、模型的有效性检验、预测能力的检验，研究结果表明：Logistic逐步回归分析得到的回归模型性能良好，回判估计的误判率为零，预测精度高。相比距离判别分析模型，回归分析模型在现场岩体分级更加方便，回判的误判率更低，另外模型能输出岩体属于各级别的概率，为工程设计人员提供更多的岩体质量信息；相比普通的回归分析，多分类有序因变量 Logistic回归更适于响应变量为有序多类别的岩体分级问题，因而岩体分级的多分类有序因变量回归模型是一种更优的岩体分级方法。

Multi Category Ordered Dependent Variable Logistic Regression Model for Rock Mass Classification

Multi-category ordered-dependent-variable Logistic regression model was introduced into the rock mass classification, rock uniaxial compressive strength, rock acoustic wave velocity, intensity of jointing, joint roughness coefficient, weathering variation coefficient of joint surface and permeability coefficient were used as independent variables, rock mass level was used as dependent variable, rock mass samples data was used to establish rock mass classification formula. Goodness of fit, model validity and predictive ability test were carried out to evaluate the correctness of the model, the results show that: Logistic regression analysis model has good performance, misjudging rate of training samples is zero, the predictive ability is strong. Compared to the distance discriminant analysis model, regression analysis model of rock mass classification is more convenient in use at site and has lower misjudging rate; besides, it can output the probability of all levels that rock mass belong to, which provide additional information of rock mass to engineering designer; compared to ordinary regression analysis, multi-category ordered-dependent-variable Logistic regression is more appropriate to analyze rock mass classification problems due to their discrete ordered dependent variables, thus multi-category ordered-dependent-variable regression model is a more superior rock mass classification method.