1\name{FrF2-package} 2\alias{FrF2-package} 3\docType{package} 4\title{ 5Fractional Factorial designs with 2-level factors 6} 7\description{ 8This package creates regular and non-regular Fractional Factorial 2-level designs. 9Furthermore, analysis tools for Fractional Factorial designs with 2-level factors 10are offered (main effects and interaction plots for all factors simultaneously, 11cube plot for looking at the simultaneous effects of three factors, full or half normal plot, 12alias structure in a more readable format than with the built-in function alias). 13 14The package works together with packages DoE.base and DoE.wrapper. 15} 16\details{ 17The package is still subject to development; most key functionality is now included. 18Please contact me, if you have suggestions. 19 20This package designs and analyses Fractional Factorial experiments with 2-level factors. 21Regular (function \code{\link{FrF2}}) and non-regular (function \code{\link{pb}}) 2-level fractional factorial 22designs can be generated. 23Function \code{\link{FrF2}}) can generate regular fractional factorials as blocked or split-plot designs, and hard-to-change 24factors can be specified in order to keep the number of level changes low. 25Regular resolution V designs larger than those obtainable from function \code{FrF2} 26can be created by function \code{\link{FrF2Large}} (these are not guaranteed to be optimal). 27Analysis facilities work for completely aliased designs only, 28i.e. e.g. not for analysing Plackett-Burman designs with interactions. 29 30Functions \code{fac.design}, \code{fractionate} or \code{oa.design} from 31Chambers and Hastie (1993) have been used as role models e.g. for 32the option \code{factor.names} or for outputting a data frame with attributes. 33However, S compatibility has not been considered in devising this package. The original 34abovementioned functions are not available in \code{R}; similar 35functions have been from package \pkg{\link[DoE.base:DoE.base-package]{DoE.base}} are 36available together with other general functionality for experimental designs. 37 38In terms of analysis, package \code{FrF2} works on linear models and enables convenient main effects and 39interaction plots (functions \code{MEPlot} and \code{IAPlot}) similar to those 40offered by Minitab software for all factors simultaneously, even though especially the 41interactions are often aliased, i.e. the model is typically singular. 42For the (less frequent) case of suspected three-factor-interactions, function 43\code{cubePlot} displays a cube with corners labeled with the (modeled) 44means of three factors simultaneously. 45Furthermore, the function \code{DanielPlot} from package \pkg{BsMD} has been 46modified to automatically label effects significant according to the 47Lenth-criterion, and to provide more usage comfort to the analyst. 48Finally, the function \code{aliases} determines the alias structure of a 49Fractional Factorial 2-level design in a format more suitable for human readers 50than the output from the built-in function \code{alias}. 51 52} 53\author{ 54Ulrike Groemping 55 56Maintainer: Ulrike Groemping <groemping@bht-berlin.de> 57} 58\references{ 59 Box G. E. P, Hunter, W. C. and Hunter, J. S. (2005) 60 \emph{Statistics for Experimenters, 2nd edition}. 61 New York: Wiley. 62 63 Chambers, J.M. and Hastie, T.J. (1993). \emph{Statistical Models in S}, 64 Chapman and Hall, London. 65 66 Chen, J., Sun, D.X. and Wu, C.F.J. (1993) 67 A catalogue of 2-level and 3-level orthogonal arrays. 68 \emph{International Statistical Review} \bold{61}, 131-145. 69 70 Daniel, C. (1959) 71 Use of Half Normal Plots in Interpreting Two Level Experiments. 72 \emph{Technometrics}, \bold{1}, 311-340. 73 74 Groemping, U. (2014). {R} Package {FrF2} for Creating and Analyzing Fractional 75 Factorial 2-Level Designs. \emph{Journal of Statistical Software}, \bold{56}, 76 Issue 1, 1-56. \url{http://www.jstatsoft.org/v56/i01/}. 77 78 Hedayat, A.S., Sloane, N.J.A. and Stufken, J. (1999) 79 \emph{Orthogonal Arrays: Theory and Applications}, Springer, New York. 80 81 Lenth, R.V. (1989) Quick and easy analysis of unreplicated factorials. 82 \emph{Technometrics}, \bold{31}, 469-473. 83 84 Mee, R. (2009). \emph{A Comprehensive Guide to Factorial Two-Level Experimentation}. 85 New York: Springer. 86 87 Montgomery, D.C. (2001). Design and Analysis of Experiments (5th ed.). Wiley, New York. 88 89 Plackett, R.L.; Burman, J.P. (1946) The design of optimum multifactorial 90 experiments. \emph{Biometrika} \bold{33}, 305-325. 91 92 Ryan, K.J. and Bulutoglu, D.A. (2010). Minimum Aberration Fractional Factorial Designs With Large N. 93 \emph{Technometrics} \bold{52}, 250-255. 94 95 Sanchez, S.M. and Sanchez, P.J. (2005). Very Large Fractional Factorial 96 and Central Composite Designs. 97 \emph{ACM Transactions on Modeling and Computer Simulation} 98 \bold{15}, 362-377. 99} 100\keyword{ array } 101\keyword{ design } 102\seealso{ 103The key design generating functions: \code{\link{FrF2}}, \code{\link{pb}}, \code{\link{FrF2Large}}\cr 104S3 class \code{\link[DoE.base:class-design]{design}}\cr 105Related packages: 106\code{\link[DoE.base:DoE.base-package]{DoE.base}}, 107\code{\link[DoE.wrapper:DoE.wrapper-package]{DoE.wrapper}}, 108\code{\link[BsMD:BsMD-package]{BsMD}};\cr 109Graphical analysis functions: \code{\link{MEPlot}}, \code{\link{IAPlot}}, \code{\link{cubePlot}}, 110\code{\link{DanielPlot}}\cr 111Analysis of alias structure for linear models of \code{FrF2} designs: \code{\link{aliases}}\cr 112} 113\examples{ 114 ### for examples on design generation, cf. functions pb and FrF2 115 116 ### Injection Molding Experiment. Box et al. 1978. 117 data(BM93.e3.data, package="BsMD") #from BsMD 118 iMdat <- BM93.e3.data[1:16,2:10] #only original experiment 119 # make data more user-friendly 120 colnames(iMdat) <- c("MoldTemp","Moisture","HoldPress","CavityThick","BoostPress", 121 "CycleTime","GateSize","ScrewSpeed", "y") 122 # linear model with all main effects and 2-factor interactions 123 iM.lm <- lm(y ~ (.)^2, data = iMdat) 124 # determine aliases 125 aliases(iM.lm) 126 # coded version 127 aliases(iM.lm, code=TRUE) 128 # normal plot of effects, default is autolabel with alpha=0.05 129 DanielPlot(iM.lm) 130 DanielPlot(iM.lm,code=TRUE) 131 DanielPlot(iM.lm,code=TRUE,alpha=0.5) 132 # half normal plot of effects 133 DanielPlot(iM.lm,code=TRUE,alpha=0.5,half=TRUE) 134 # main effects plots 135 MEPlot(iM.lm) 136 # interaction plots 137 IAPlot(iM.lm) 138 # interaction plots with attention drawn to aliases 139 aus <- IAPlot(iM.lm, show.alias=TRUE) 140 # alias groups corresponding to interaction plots 141 aliases(iM.lm)$aliases[9:15] 142 # returned object 143 aus 144 # cube plot of three factors 145 # (not very useful for this model, for demonstration only) 146 ## per default, modeled means are shown 147 ## this does not make a difference here, since the main effect of 148 ## ScrewSpeed is confounded with the MoldTemp:HoldPress:BoostPress 149 ## interaction, so that the three-factor-interaction is indirectly included 150 ## in the modeled means 151 cubePlot(iM.lm, "MoldTemp", "HoldPress", "BoostPress") 152 ## modeled means without a three-factor interaction 153 cubePlot(lm(y ~ (MoldTemp+HoldPress+BoostPress)^2, data = iMdat), 154 "MoldTemp", "HoldPress", "BoostPress") 155 ## modeled=FALSE reverts to showing the apparent three-factor interaction 156 cubePlot(lm(y ~ (MoldTemp+HoldPress+BoostPress)^2, data = iMdat), 157 "MoldTemp", "HoldPress", "BoostPress", modeled=FALSE) 158 ## cubePlot also works on raw data 159 cubePlot(iMdat$y, iMdat$MoldTemp, iMdat$HoldPress, iMdat$BoostPress) 160 ## plotting functions also work directly on designs, 161 ## if these have been generated from functions FrF2 or pb: 162 plan <- FrF2(16, 7) 163 plan <- add.response(plan, rnorm(16)) 164 MEPlot(plan) 165 IAPlot(plan) 166 DanielPlot(plan) 167 168} 169