rscanon(y,x1,x2,...,xk [,block:var1,block:var2,...]), REAL vectors y, x1, ...,xk, factors var1, var2, ...; all should have the same number of rows. |

rscanon(y,x1,x2,...,xk) performs the canonical analysis for the quadratic response surface model with response y and predictors x1, ... , xk. y and x1 through xk must be REAL vectors of the same length. The result is a structure with components 'b0', 'b', 'B', 'x0', 'y0', 'H', and 'lambda', giving the intercept, linear coefficients, the quadratic/ cross product coefficient matrix, the stationary point, the predicted response at the stationary point, the matrix of canonical directions, and the eigenvalues, respectively. If the design was blocked, you can specify the blocking factors using one or more keyword phrases of the form 'block:var', where var is a factor. For example, if the design was blocked by factors date and analyst, you might use Cmd> rscanon(yield,time,temperature,block:date,block:analyst). The output is the same as before, but is block adjusted. Examples: Cmd> #data from example 16-2 of Montgomery Cmd> x1 <- vector(-1,-1,1,1,0,0,0,0,0,1.414,-1.414,0,0) Cmd> x2 <- vector(-1,1,-1,1,0,0,0,0,0,0,0,1.414,-1.414) Cmd> y <- vector(76.5,77.0,78.0,79.5,79.9,80.3,80.0,79.7,\ 79.8,78.4,75.6,78.5,77.0) Cmd> rscanon(y,x1,x2) component: b0 (1) 79.94 component: b (1) 0.99505 0.5152 component: B (1,1) -1.3764 0.125 (2,1) 0.125 -1.0013 component: x0 (1) 0.38923 0.30585 component: y0 (1) 80.212 component: H (1,1) 0.28972 0.95711 (2,1) 0.95711 -0.28972 component: lambda (1) -0.9635 -1.4143

Gary Oehlert 2003-01-15