version 1.1

cran · Sep 5, 2016 · 663692f · 663692f
1 parent acf55ed
commit 663692f
Show file tree

Hide file tree

Showing 19 changed files with 574 additions and 299 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,13 +1,21 @@
 Package: MCDM
-Title: Multi-Criteria Decision Making Methods
-Version: 1.0
-Date: 2015-09-10
+Type: Package
+Title: Multi-Criteria Decision Making Methods for Crisp Data
+Version: 1.1
+Date: 2016-09-05
 Author: Blanca A. Ceballos Martin <[email protected]>
 Maintainer: Blanca A. Ceballos Martin <[email protected]>
-Description: An R implementation of Four Multi-Criteria Decision Making (MCDM) Methods: Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), "VIseKriterijumska Optimizacija I Kompromisno Resenje" (VIKOR), both Multi-Objective Optimization by Ratio Analysis and Full Multiplicative Form (Multi-MOORA) and Weighted Aggregated Sum Product ASsessment (WASPAS). In addition, this package provides a MetaRanking function which combines the output of the previous methods. 
+Description: Implementation of several MCDM methos for crisp data for decision
+    making problems. The methods that are implemented in this package are RIM,
+    TOPSIS (with two normalization procedures), VIKOR, Multi-MOORA and WASPAS.
+    In addition, MetaRanking function calculates a new ranking from the sum 
+    of the rankings calculated, as well as an aggregated ranking.
+Imports: RankAggreg
 License: LGPL (>= 3)
 URL: http://decsai.ugr.es/index.php?p=miembros&id=19909
+LazyData: true
+RoxygenNote: 5.0.1
 NeedsCompilation: no
-Packaged: 2015-11-18 16:00:54 UTC; Blanky
+Packaged: 2016-09-05 11:31:04 UTC; Modo
 Repository: CRAN
-Date/Publication: 2015-11-19 11:02:12
+Date/Publication: 2016-09-05 13:53:38
diff --git a/MD5 b/MD5
@@ -1,13 +1,17 @@
-b301612e520e87ecf367a63d632b0645 *DESCRIPTION
-0f714634d9b6e945d6b30e5d6276b22e *NAMESPACE
-a6d7468cec7f23a7d52226beb220a609 *R/MMOORA.R
-916b563fb195602c830dd3c55b3f1f7e *R/MetaRanking.R
-5b246b44eab588a3641bf7288c90f566 *R/TOPSIS.R
-6f0bdad34dcfb7883f3dc08ebdecb72e *R/VIKOR.R
-f80bef4b19a66218360122eb29b1f0b0 *R/WASPAS.R
-e91d631037e23f00c7fd70739a1278c3 *man/MMOORA.Rd
-c3d685827adbeb13256720e443d2687a *man/MetaRanking.Rd
-a263ae3ec5866104e5736868d3ebbd8e *man/TOPSIS.Rd
-73dae3c80f82ad4dd9119924e75ce00d *man/VIKOR.Rd
-244851b5223807a97aaceba704f70a8e *man/WASPAS.Rd
-9120b7e76ff718096df3feea7dce908d *vignettes/MCDM.pdf
+69910b71bb5bd011e994234c8b59b0c1 *DESCRIPTION
+9b7c4e38c891101fd4549788a687c7de *NAMESPACE
+8287ca47154ab5e73e30274a16880212 *R/MMOORA.R
+5b035e73afcd7f7bae9d167620a5a93b *R/MetaRanking.R
+76fbfe56c0a7f78ed63313786b5cff85 *R/RIM.R
+00ac3bdfc38bc0f08b8c3c52233e6555 *R/TOPSISLinear.R
+4688cfc2894c6138351b9d12525d697d *R/TOPSISVector.R
+8d25956de81910c4f10a7d01606915a3 *R/TheoryOfDominance.R
+36138b0987e910975df5fd5e956556a2 *R/VIKOR.R
+b5f8ffd3b9c728bd73b802be03486e88 *R/WASPAS.R
+6803bdbf53ad9f3fd94ebe327a580fb6 *man/MMOORA.Rd
+7e39cee422906ca0e238c31f6aaa3d15 *man/MetaRanking.Rd
+cbd050e6512a7d2e1db0dfb7c996371d *man/RIM.Rd
+2cb23006a72a752f685f0e05cd52e4f7 *man/TOPSISLinear.Rd
+f757fbf818251edabcb38bdf2d81879f *man/TOPSISVector.Rd
+33a7b42ebabfc5a90616c496db7639c4 *man/VIKOR.Rd
+3369c5e7b9317128957225d4a03fd52b *man/WASPAS.Rd
diff --git a/NAMESPACE b/NAMESPACE
@@ -1,2 +1,3 @@
-exportPattern("^[[:alpha:]]+")
-
+import(RankAggreg)
+export(MetaRanking, MMOORA, RIM, TOPSISVector, TOPSISLinear, VIKOR, WASPAS)
+
diff --git a/R/MMOORA.R b/R/MMOORA.R
@@ -1,16 +1,18 @@
 #' Implementation of MULTIMOORA Method for Multi-Criteria Decision Making Problems.
 #'
 #' @description The \code{MMOORA} function implements both the Multi-Objetive Optimization by Ration Analysis (MOORA) and the "Full Multiplicative Form" (MULTIMOORA).
-#' @param decision The decision matrix (\emph{m} x \emph{n}) with the values of the \emph{m} alternatives, for the \emph{n} criteria. 
+#' @param decision The decision matrix (\emph{m} x \emph{n}) with the values of the \emph{m} alternatives, for the \emph{n} criteria.
 #' @param weights A vector of length \emph{n}, containing the weights for the criteria. The sum of the weights has to be 1.
 #' @param cb A vector of length \emph{n}. Each component is either \code{cb(i)='max'} if the \emph{i-th} criterion is benefit or \code{cb(i)='min'} if the \emph{i-th} criterion is a cost.
-#' @return \code{MMOORA} returns a data frame which contains the scores and the four rankings calculated (Ratio System, Reference Point, Multiplicative Form and Multi-MOORA ranking). 
+#' @return \code{MMOORA} returns a data frame which contains the scores and the four rankings calculated (Ratio System, Reference Point, Multiplicative Form and Multi-MOORA ranking).
 #' @references Brauers, W. K. M.; Zavadskas, E. K. Project management by MULTIMOORA as an instrument for transition economies. Technological and Economic Development of Economy, 16(1), 5-24, 2010.
 #' @examples
-#' 
-#'  d <- matrix(rpois(12, 5), nrow = 4)
-#'  w <- c(0.2, 0.2, 0.6)
-#'  cb <- c('max','min','max')
+#'
+#'  d <- matrix(c(60,6.35,6.8,10,2.5,4.5,3,0.4,0.15,0.1,0.2,0.1,0.08,0.1,2540,1016,1727.2,
+#'  1000,560,1016,1778,500,3000,1500,2000,500,350,1000,990,1041,1676,965,915,508,920),
+#'  nrow=7,ncol=5)
+#'  w <- c(0.036,0.192,0.326,0.326,0.12)
+#'  cb <- c('max','min','max','max','max')
 #'  MMOORA(d,w,cb)
 
 MMOORA <- function(decision, #matrix with all the alternatives
@@ -33,10 +35,10 @@ MMOORA <- function(decision, #matrix with all the alternatives
     stop("length of 'weights' does not match the number of the criteria")
   if(length(cb) != ncol(decision))
     stop("length of 'cb' does not match the number of the criteria")
-  
-  
+
+
   #MMOORA method
-  
+
   #1. Normalization and weighting
   d = sqrt(colSums(decision^2))
   NW <- matrix(nrow = nrow(decision), ncol = ncol(decision))
@@ -48,44 +50,39 @@ MMOORA <- function(decision, #matrix with all the alternatives
   NR <- NW
   for(j in 1:ncol(decision)){
     if (cb[j] == 'min'){
-      NR[,j] <- NW[,j]*(-1)    
+      NR[,j] <- NW[,j]*(-1)
     }
   }
   RS <- apply(NR, 1, sum)
-  
+
   #3. Reference point
-  Ref <- as.integer(cb == "max") * apply(NW, 2, max) + 
+  Ref <- as.integer(cb == "max") * apply(NW, 2, max) +
     as.integer(cb == "min") * apply(NW, 2, min)
   RefP <- matrix(nrow = nrow(decision), ncol = ncol(decision))
   for(j in 1:ncol(decision)){
     RefP[,j] <- abs(Ref[j]-NW[,j])
   }
   RP <- apply(RefP, 1, max)
-  
+
   #4. Multiplicative form
-  NEW <- matrix(nrow = nrow(decision), ncol = ncol(decision))
-  for(j in 1:ncol(decision)){
-    NEW[,j] <- (decision[,j] / d[j]) ^ weights[j]
-  }
-
-  max <- NEW
-  min <- NEW
-  for (j in 1:ncol(NEW)){
+  max <- NW
+  min <- NW
+  for (j in 1:ncol(NW)){
     if (cb[j] == 'max'){
       min[,j] <- 1
     }else{
       max[,j] <- 1
-    } 
+    }
   }
   A <- apply(max, 1, prod)
   B <- apply(min, 1, prod)
   M <- A/B
 
   #5. Ranking the alternatives
-  Rrs <- rank(-RS)
-  Rrp <- rank(RP)
-  Rm <- rank(-M)
-  R <- Rrs + Rrp + Rm
-  return(data.frame(Alternatives = 1:nrow(decision), RatioSystem = RS, Ranking = rank(-RS, ties.method= "random"), ReferencePoint = RP, Ranking = rank(RP, ties.method= "random"), MultiplicativeForm = M, Ranking = rank(-M, ties.method= "random"), MultiMooraRanking = rank(R, ties.method= "random")))
-  
-}
+  Rrs <- rank(-RS, ties.method= "first")
+  Rrp <- rank(RP, ties.method= "first")
+  Rm <- rank(-M, ties.method= "first")
+  MMRanking = TheoryOfDominance(Rrs,Rrp,Rm,decision)
+  return(data.frame(Alternatives = 1:nrow(decision), RatioSystem = RS, Ranking = Rrs, ReferencePoint = RP, Ranking = Rrp, MultiplicativeForm = M, Ranking = Rm, MultiMooraRanking = MMRanking))
+
+}
diff --git a/R/MetaRanking.R b/R/MetaRanking.R
@@ -1,63 +1,64 @@
-#' Implementation of MetaRanking function for Multi-Criteria Decision Making Problems.
-#'
-#' @description The \code{MetaRanking} function internally calls functions  \code{MMOORA}, \code{TOPSIS}, \code{VIKOR} and \code{WASPAS} and then calculates a sum of the their rankings.
-#' @param decision The decision matrix (\emph{m} x \emph{n}) with the values of the \emph{m} alternatives, for the \emph{n} criteria. 
-#' @param weights A vector of length \emph{n}, containing the weights for the criteria. The sum of the weights has to be 1.  
-#' @param cb A vector of length \emph{n}. Each component is either \code{cb(i)='max'} if the \emph{i-th} criterion is benefit or \code{cb(i)='min'} if the \emph{i-th} criterion is a cost.
-#' @param lambda A value in [0,1]. It is used in the calculation of the W index for WASPAS method.
-#' @param v A value in [0,1]. It is used in the calculation of the Q index for VIKOR method.
-#' @return \code{MetaRanking} returns a data frame which contains the rankings of the Multi-MOORA, TOPSIS, VIKOR, WASPAS Methods and the MetaRanking of the alternatives. 
-#' @examples
-#' 
-#'  d <- matrix(rpois(12, 5), nrow = 4)
-#'  w <- c(0.2, 0.2, 0.6)
-#'  cb <- c('max','min','max')
-#'  lambda <- 0.5
-#'  v <- 0.5
-#'  MetaRanking(d,w,cb,lambda,v)
-
-MetaRanking <- function(decision, #matrix with all the alternatives
-                        weights,  #vector with the numeric values of the weights
-                        cb,       #vector with the "type" of the criteria (benefit = "max", cost = "min")
-                        lambda,   #value with the real number of the 'lambda' parameter to calculate W
-                        v         #value with the real number of the 'v' parameter to calculate Q
-)
-{
-  #Checking the arguments
-  if(! is.matrix(decision))
-    stop("'decision' must be a matrix with the values of the alternatives")
-  if(missing(weights))
-    stop("a vector containing n weigths, adding up to 1, should be provided")
-  if(sum(weights) != 1)
-    stop("The sum of 'weights' is not equal to 1")
-  if(! is.character(cb))
-    stop("'cb' must be a character vector with the type of the criteria")
-  if(! all(cb == "max" | cb == "min"))
-    stop("'cb' should contain only 'max' or 'min'")
-  if(length(weights) != ncol(decision))
-    stop("length of 'weights' does not match the number of the criteria")
-  if(length(cb) != ncol(decision))
-    stop("length of 'cb' does not match the number of the criteria")
-  if(missing(lambda))
-    stop("a value for 'lambda' in [0,1] should be provided")
-  if(missing(v))
-    stop("a value for 'v' in [0,1] should be provided")
-
-
-  #Multi-MOORA method 
-  MMoora = MMOORA(decision,weights,cb)
-
-  #TOPSIS method 
-  Topsis = TOPSIS(decision,weights,cb)
-
-  #VIKOR method  
-  Vikor = VIKOR(decision,weights,cb,v)
-
-  #WASPAS method  
-  Waspas = WASPAS(decision,weights,cb,lambda)
-
-  #Meta-Ranking
-  MetaR = MMoora[,8]+Topsis[,3]+Vikor[,5]+Waspas[,3]
-  return(data.frame(Alternatives = 1:nrow(decision), MMOORA = MMoora[,8], TOPSIS = Topsis[,3], VIKOR = Vikor[,5], WASPAS = Waspas[,3], METARANKING = rank(MetaR, ties.method= "random")))
-
-}
+#' Implementation of MetaRanking function for Multi-Criteria Decision Making Problems.
+#'
+#' @description The \code{MetaRanking} function internally calls functions  \code{MMOORA}, \code{RIM}, \code{TOPSISLinear}, \code{TOPSISVector}, \code{VIKOR} and \code{WASPAS} and then calculates a sum of the their rankings and an aggregated ranking by applying the \code{RankAggreg} package.
+#' @param decision The decision matrix (\emph{m} x \emph{n}) with the values of the \emph{m} alternatives, for the \emph{n} criteria.
+#' @param weights A vector of length \emph{n}, containing the weights for the criteria. The sum of the weights has to be 1.
+#' @param cb A vector of length \emph{n}. Each component is either \code{cb(i)='max'} if the \emph{i-th} criterion is benefit or \code{cb(i)='min'} if the \emph{i-th} criterion is a cost.
+#' @param lambda A value in [0,1]. It is used in the calculation of the W index for WASPAS method.
+#' @param v A value in [0,1]. It is used in the calculation of the Q index for VIKOR method.
+#' @param AB A matrix (\emph{2} x \emph{n}). AB[1,] corresponds with the A extrem, and AB[2,] represents the B extrem of the domain of each criterion.
+#' @param CD A matrix (\emph{2} x \emph{n}). CD[1,] corresponds with the C extrem, and CD[2,] represents the D extrem of the ideal reference of each criterion.
+#' @return \code{MetaRanking} returns a data frame which contains the rankings of the Multi-MOORA, RIM, TOPSISLinear, TOPSISVector, VIKOR, WASPAS Methods and the both MetaRankings of the alternatives.
+#' @examples
+#'
+#'  d <- matrix(c(1,2,5,3000,3750,4500),nrow = 3,ncol = 2)
+#'  w <- c(0.5,0.5)
+#'  cb <- c('min','max')
+#'  lambda <- 0.5
+#'  v <- 0.5
+#'  AB <- matrix(c(1,5,3000,4500),nrow = 2,ncol=2)
+#'  CD <- matrix(c(1,1,4500,4500),nrow = 2,ncol=2)
+#'  MetaRanking(d,w,cb,lambda,v,AB,CD)
+
+MetaRanking <- function(decision, #matrix with all the alternatives
+                        weights,  #vector with the numeric values of the weights
+                        cb,       #vector with the "type" of the criteria (benefit = "max", cost = "min")
+                        lambda,   #value with the real number of the 'lambda' parameter to calculate W
+                        v,       #value with the real number of the 'v' parameter to calculate Q
+                        AB, #matrix with the range [A,B] of the universe of discourse
+                        CD #matrix with the Reference Ideal [C,D]
+)
+{
+
+  #Multi-MOORA method
+  MMoora = MMOORA(decision,weights,cb)
+
+  #RIM
+  Rim = RIM(decision, weights, AB, CD)
+
+  #TOPSIS method
+  TopsisV = TOPSISVector(decision,weights,cb)
+  TopsisL = TOPSISLinear(decision,weights,cb)
+
+  #VIKOR method
+  Vikor = VIKOR(decision,weights,cb,v)
+
+  #WASPAS method
+  Waspas = WASPAS(decision,weights,cb,lambda)
+
+  #Meta-Ranking
+  MetaR = MMoora[,8]+Rim[,3]+TopsisV[,3]+TopsisL[,3]+Vikor[,5]+Waspas[,5]
+
+  #Ranking Aggregated
+  #library(RankAggreg)
+  ra = rbind(MMoora[,8],Rim[,3],TopsisV[,3],TopsisL[,3],Vikor[,5],Waspas[,5])
+  if(nrow(decision)<=10)
+    RA = RankAggreg::BruteAggreg(ra, nrow(decision), distance="Spearman")
+  else
+    RA = RankAggreg::RankAggreg(ra, nrow(decision), method = "GA", distance = "Spearman", verbose=FALSE)
+
+  return(data.frame(Alternatives = 1:nrow(decision), MMOORA = MMoora[,8], RIM = Rim[,3], TOPSISVector = TopsisV[,3],
+                    TOPSISLinear = TopsisL[,3], VIKOR = Vikor[,5], WASPAS = Waspas[,5],
+                    MetaRanking_Sum = rank(MetaR, ties.method= "first"), MetaRanking_Aggreg = RA$top.list))
+
+}
diff --git a/R/RIM.R b/R/RIM.R
@@ -0,0 +1,76 @@
+#' Implementation of RIM Method for Multi-Criteria Decision Making Problems.
+#'
+#' @description The \code{RIM} function implements the Reference Ideal Method (RIM).
+#' @param decision The decision matrix (\emph{m} x \emph{n}) with the values of the \emph{m} alternatives, for the \emph{n} criteria.
+#' @param weights A vector of length \emph{n}, containing the weights for the criteria. The sum of the weights has to be 1.
+#' @param AB A matrix (\emph{2} x \emph{n}). AB[1,] corresponds with the A extrem, and AB[2,] represents the B extrem of the domain of each criterion.
+#' @param CD A matrix (\emph{2} x \emph{n}). CD[1,] corresponds with the C extrem, and CD[2,] represents the D extrem of the ideal reference of each criterion.
+#' @return \code{RIM} returns a data frame which contains the score of the R index and the ranking of the alternatives.
+#' @references Cables, E.; Lamata, M.T.; Verdegay, J.L. RIM-reference ideal method in multicriteria decision making. Information Science, 337-338, 1-10, 2016.
+#' @examples
+#'
+#'  d <- matrix(c(30,40,25,27,45,0,9,0,0,15,2,1,3,5,2,3,3,1,3,2,3,2,3,3,3,2,2,2,1,4),
+#'  nrow = 5, ncol = 6)
+#'  w <- c(0.2262,0.2143,0.1786,0.1429,0.119,0.119)
+#'  AB = matrix(c(23,60,0,15,0,10,1,3,1,3,1,5),nrow = 2,ncol = 6)
+#'  CD = matrix(c(30,35,10,15,0,0,3,3,3,3,4,5),nrow = 2,ncol = 6)
+#'  RIM(d,w,AB,CD)
+
+RIM <- function(decision, #matrix with all the alternatives
+                weights,  #vector with the numeric values of the weights
+                AB, #matrix with the range [A,B] of the universe of discourse
+                CD #matrix with the Reference Ideal [C,D]
+)
+{
+  #Checking the arguments
+  if(! is.matrix(decision))
+    stop("'decision' must be a matrix with the values of the alternatives")
+  if(missing(weights))
+    stop("a vector containing n weigths, adding up to 1, should be provided")
+  if(sum(weights) != 1)
+    stop("The sum of 'weights' is not equal to 1")
+  if(length(weights) != ncol(decision))
+    stop("length of 'weights' does not match the number of the criteria")
+  if(ncol(AB) != ncol(decision))
+    stop("length of 'AB' does not match the number of the criteria")
+  if(ncol(CD) != ncol(decision))
+    stop("length of 'CD' does not match the number of the criteria")
+
+
+  #1. Normalization and weighting
+  N <- matrix(nrow = nrow(decision), ncol = ncol(decision))
+  for(j in 1:ncol(decision)){
+    for(i in 1:nrow(decision)){
+      if((decision[i,j] >= CD[1,j]) && (decision[i,j]<= CD[2,j])){
+        N[i,j]=1
+      }
+      else if( ((decision[i,j]>= AB[1,j]) && (decision[i,j]<= CD[1,j])) && (AB[1,j]!=CD[1,j])){
+        N[i,j]=1-(min(abs(decision[i,j]-CD[1,j]),abs(decision[i,j]-CD[2,j]))/abs(AB[1,j]-CD[1,j]))
+      }
+      else if( ((decision[i,j]>= CD[2,j]) && (decision[i,j]<= AB[2,j])) && (CD[2,j]!=AB[2,j])){
+        N[i,j]=1-(min(abs(decision[i,j]-CD[1,j]),abs(decision[i,j]-CD[2,j]))/abs(CD[2,j]-AB[2,j]))
+      }
+      else stop("error in normalization procedure: A != B and D != C")
+    }
+  }
+
+
+  W <- diag(weights)
+  NW <- N%*%W
+
+  #2. Distances to the ideal solutions
+  posDis = c(1:nrow(decision))
+  negDis = c(1:nrow(decision))
+  for(i in 1:nrow(decision)){
+    posDis[i] = sqrt(sum((NW[i,]-weights)^2))
+    negDis[i] = sqrt(sum(NW[i,]^2))
+  }
+
+
+  #4. R index
+  R <- negDis/(negDis+posDis)
+
+  #5. Rank the alternatives
+  return(data.frame(Alternatives = 1:nrow(decision), R = R, Ranking = rank(-R, ties.method= "first")))
+
+}