【摘要】 系统形变往往由很多主导因素引起,且各主导因素之间并不独立。以国土利用变化驱动分析为例,运用偏最小二乘回归方法建立变化的驱动机制模型,并和主成分回归法相比较。结果表明,偏最小二乘回归不仅在拟合效果上优于主成分回归,系统性更强,结论更可靠,而且偏最小二乘模型的回归系数更易于解释,提供的系统信息也更丰富。偏最小二乘为样本个数少、自变量多、且变量间存在多重共线性的复杂大系统形变分析提供了新的、有效的解决途径。更多还原

【Abstract】 System deformation is usually caused by many leading factors,which aren’t independent.Taking driving analysis of the land use change as an example,a dynamic models is set up based on partial least squares regression.and also compared with principal component regression.The results show that partial least squares regression not only has better fitting,stronger systematic and reliable than principal component regression,but also the coefficients are easily explicated and much large systematic information are transmitted.Partial least squares will provide a new and effective analysis method for complicated and big system which has less samples,more independent and multicollinearity variables.更多还原