线性判别分析(LDA分析)及相关R语言code.pdfVIP

Introduction to Linear Discriminant Analysis Example Using an Untrained LDA Model Lecture 13 - Winter 2011 February 23, 2011 Lecture 13 Introduction to Linear Discriminant Analysis Basic Linear Discriminant Model • LDA computes linear combinations of variables based on the following equation: D = d z + d z + . . . + d z i i 1 i 1 i2 i2 ik ik where: Di = the discriminant functions (i = # groups – 1) dik = standardized discriminant coefficients (k = # variables) zik = standardized variable values (e.g., standardized sepal lengths) • LDA is informed (initial group membership must be specified) • The coefficients are chosen to maximize separation between groups • LDA is based on a linear model, and assumes normality and homoscedasticity • LDA can omit variables that don’t contribute to the linear model Lecture 13 Introduction to Linear Discriminant Analysis Using the Linear Discriminant Model to Classify Samples • The discriminant model is used to generate a classification equation: C = c + c X + c X + . . . + c X g g 0 g 1 g 1 g 2 g 2 gk gk where: C = the sample classification score for each group (g ) g cg 0 = a constant (similar to the intercept) cgk = classificati