Blend {Blendstat} | R Documentation |
Joint analysis of experiments with mixtures and random effects, taking on a process variable represented by a covariable.
Blend(Exp, X, Y, Conc = NULL, Effects = NULL)
Exp |
Vector with the names of the experiments. |
X |
Mixture variables (components), without the vector of the concentrations (covariable). |
Y |
Response variable. |
Conc |
Vector with the concentrations (covariable) of the experiments. |
Effects |
Vector of the effects of the mixtures in a reference mixture (example: centroid). |
MPred |
Matrix with the predicted and observed values. |
MCPred |
Matrix with the values predicted by components. |
MExp |
Matrix with the design of the experiments. |
Theta |
Vector with the Theta estimates. |
Marcelo Angelo Cirillo
Paulo Cesar Ossani
Kalirajan, K. P.. On the estimation of a regression model with fixed and random coefficients. Journal of Applied Statistics, 17(2): 237-244, 1990. doi:10.1080/757582835
Swany, P. A. V. B. Statistical Inference in Random Coefficient Regression Models. Amsterdam: Springer Science & Business Media, 1971. 209 p.
data(DataNAT) # dataset Exp <- DataNAT[,2] # identification of experiments X <- DataNAT[,3:6] # independent variable Y <- DataNAT[,11] # dependent variable # effects of the blends in a reference mixture Effects <- rep(c(-0.1,0,0.1,0.2,0.3,0.4,0.5,0.6,0.7),4) Conc <- as.matrix(DataNAT[,7]) # covariate (process variable) Res <- Blend(Exp = Exp, X = X, Y = Y, Conc = Conc, Effects = Effects) print("Predicted and observed values"); Res$MPred print("Values predicted by components:"); Res$MCPred print("Design of the experiments:"); Res$MExp print("Estimates of the linear model parameters:"); Res$Theta Tit <- c("Covariate (process variable)","Variable") Xlab = "Effects" # label of the X axis Ylab = "Predicted values" # label of the Y axis Plot.Blend(Res, Titles = Tit, PosLeg = 2, xlabel = Xlab, ylabel = Ylab, BoxLeg = TRUE, Color = TRUE, Casc = TRUE)