R/optionclasses.R

In blockcluster: Co-Clustering Package for Binary, Categorical, Contingency and Continuous Data-Sets

#'@include coclusterStrategy.R
NULL

#' Common Input/Output options.
#'
#' This class contains all the input options and common output options for all
#' kinds of data-sets (Binary, Categorical, Contingency and Continuous).
#'
#' \describe{
#' \item{The following are the various input options:}{}
#' \item{data: }{Input data.}
#' \item{datatype: }{Type of data.}
#' \item{semisupervised: }{Boolean value specifying if Co-clustering is semi-supervised or not.}
#' \item{model: }{Model to be run for co-clustering.}
#' \item{nbcocluster: }{Number of row and column clusters.}
#' \item{strategy: }{Input strategy.}
#' \item{The following are the various common output options:}{}
#' \item{message: }{Status returned.}
#' \item{rowproportions: }{Vector of row proportions.}
#' \item{colproportions: }{Vector of column proportions.}
#' \item{rowclass: }{Vector of assigned row cluster to each row.}
#' \item{colclass: }{Vector of assigned column cluster to each column.}
#' \item{likelihood: }{Final pseudo log-likelihood.}
#' \item{rowposteriorprob: }{Final posterior probabilities for rows.}
#' \item{colposteriorprob: }{Final posterior probabilities for columns.}
#'
#' }
#'
#' @name CommonOptions-class
# @exportClass CommonOptions
setClass(
		Class = "CommonOptions",
		representation = representation(
				data              = "matrix",
        rowlabels         = "integer",
        collabels         = "integer",
				datatype          = "character",
				model             = "character",
				nbcocluster       = "numeric",
		    semisupervised    = "logical",
				strategy          = "strategy",
				message           = "character",
				successful        = "logical",
				rowproportions    = "numeric",
				columnproportions = "numeric",
				rowclass          = "numeric",
				colclass          = "numeric",
				rowposteriorprob  = "numeric",
				colposteriorprob  = "numeric",
				likelihood        = "numeric",
				coclusterdata     = "matrix",
				ICLvalue          = "numeric"
		),
		prototype = prototype(
				data              = matrix(nrow=0,ncol=0),
    		rowlabels         = integer(0),
    		collabels         = integer(0),
				datatype          = character(0),
				model             = character(0),
				nbcocluster       = numeric(0),
    		semisupervised    = logical(0),
				message           = character(0),
				successful        = logical(0),
				rowproportions    = numeric(0),
				columnproportions = numeric(0),
				rowclass          = numeric(0),
				colclass          = numeric(0),
				rowposteriorprob  = numeric(0),
				colposteriorprob  = numeric(0),
				likelihood        = numeric(0),
				coclusterdata     = matrix(nrow=0,ncol=0),
				ICLvalue          = numeric(0)
		)
)

#'
#' Binary input/output options
#'
#' This class contains all the input options as well as the estimated paramters for Binary data-set. It inherits
#' from base class \code{\linkS4class{CommonOptions}}. The class contains following output parameters given in 'Details' along
#' with the parameters in base class.
#'
#' \describe{
#' \item{classmean:}{The mean value of each co-cluster.}
#' \item{classdispersion:}{The dispersion of each co-cluster.}
#' \item{ICLvalue: }{Integrated complete likelihood}
#' }
#'
#' @name BinaryOptions-class
# @exportClass BinaryOptions
setClass(
		Class = "BinaryOptions",
		contains = "CommonOptions",
		representation = representation(
				classmean       = "matrix",
				classdispersion = "matrix",
				hyperparam      = "numeric"
		),

		prototype = prototype(
				classmean = matrix(nrow=0,ncol=0),
				classdispersion = matrix(nrow=0,ncol=0),
        hyperparam = c(1,1)
		)
)

#'
#' Contingency input/output options
#'
#' This class contains all the input options as well as the estimated paramters for Contingency data-set.It inherits
#' from base class \code{\linkS4class{CommonOptions}}. The class contains following output parameters given in 'Details' along
#' with the parameters in base class.
#'
#' \describe{
#' \item{classgamma:}{The value of poisson parameter (gamma) for each co-cluster.}
#' \item{datamui:}{Rows effect (if known).}
#' \item{datanuj:}{Columns effect (if known).}
#' }
#'
#' @name ContingencyOptions-class
# @exportClass ContingencyOptions
setClass(
		Class = "ContingencyOptions",
		contains = "CommonOptions",
		representation = representation(
				classgamma = "matrix",
				datamui    = "numeric",
				datanuj    = "numeric"
		),
		prototype = prototype(
				classgamma = matrix(nrow=0,ncol=0),
				datamui = numeric(0),
				datanuj = numeric(0)
		)
)

#' Continuous input/output options
#'
#' This class contains all the input options as well as the estimated parameters
#' for Continuous data-sets. It inherits from base class \code{\linkS4class{CommonOptions}}. The class contains following output parameters given in 'Details' along
#' with the parameters in base class.
#'
#' \describe{
#' \item{classmean:}{The mean value of each co-cluster.}
#' \item{classvariance:}{The variance of each co-cluster.}
#' }
#'
#' @name ContinuousOptions-class
# @exportClass ContinuousOptions
setClass(
		Class = "ContinuousOptions",
		contains = "CommonOptions",
		representation = representation(
				classmean     = "matrix",
				classvariance = "matrix"
		),
		prototype = prototype(
				classmean     = matrix(nrow=0,ncol=0),
				classvariance = matrix(nrow=0,ncol=0)
		)
)

#'
#' Categorical input/output options
#'
#' This class contains all the input options as well as the estimated paramters for categorical data-set. It inherits
#' from base class \code{\linkS4class{CommonOptions}}. The class contains following output parameters given in 'Details' along
#' with the parameters in base class.
#'
#' \describe{
#' \item{classmean:}{The categorical distribution of each co-cluster}
#' \item{ICLvalue: }{Integrated complete likelihood}
#' }
#'
#' @name CategoricalOptions-class
# @exportClass CategoricalOptions
setClass(
  Class = "CategoricalOptions",
  contains = "CommonOptions",
  representation = representation(
    classmean    = "list",
		hyperparam   = "numeric"
    ),

  prototype = prototype(
    classmean = list(0),
		hyperparam = c(1,1)
  )
)


#' Getter method for blockcluster output
#'
#' This is overloading of square braces to extract values of various slots of
#' the output from \code{\link{coclusterBinary}},
#' \code{\link{coclusterCategorical}}, \code{\link{coclusterContingency}},
#' \code{\link{coclusterContinuous}}.
#'
#' @param x object from which to extract element(s) or in which to replace element(s).
#' @param i the name of the element we want to extract or replace.
#' @param j if the element designing by i is complex, j specifying elements to extract or replace.
#' @param drop not used
#'
#' @name [
#' @rdname getter-methods
NULL


#' @rdname getter-methods
#' @aliases [,BinaryOptions-method
#'

setMethod(
		f = "[",
		signature = "BinaryOptions",
		definition = function(x,i,j,drop) {
			switch(EXPR=i,
					"data"={return (x@data)},
          "coclusterdata"={return (x@coclusterdata)},
					"datatype"={return (x@datatype)},
          "semisupervised"={return (x@semisupervised)},
					"model"={return (x@model)},
					"strategy"={return (x@strategy)},
					"message"={return (x@message)},
					"rowproportions"={return (x@rowproportions)},
					"colproportions"={return (x@colproportions)},
					"rowclass"={return (x@rowclass)},
					"colclass"={return (x@colclass)},
					"likelihood"={return (x@likelihood)},
					"rowposteriorprob"={return (x@rowposteriorprob)},
					"colposteriorprob"={return (x@colposteriorprob)},
					"classmean"={return (x@classmean)},
					"classdispersion"={return (x@classdispersion)},
			    "ICLvalue"={return (x@ICLvalue)},
					"hyperparam"={return (x@hyperparam)},
						stop("Invalid slot name.")
			)
		}
)

#' @rdname getter-methods
#' @aliases [,ContingencyOptions-method
#'
setMethod(
		f = "[",
		signature = "ContingencyOptions",
		definition = function(x,i,j,drop) {
			switch(EXPR=i,
					"data"={return (x@data)},
			    "coclusterdata"={return (x@coclusterdata)},
          "datatype"={return (x@datatype)},
			    "semisupervised"={return (x@semisupervised)},
					"model"={return (x@model)},
					"strategy"={return (x@strategy)},
					"message"={return (x@message)},
					"rowproportions"={return (x@rowproportions)},
					"colproportions"={return (x@colproportions)},
					"rowclass"={return (x@rowclass)},
					"colclass"={return (x@colclass)},
					"likelihood"={return (x@likelihood)},
					"rowposteriorprob"={return (x@rowposteriorprob)},
					"colposteriorprob"={return (x@colposteriorprob)},
					"classgamma"={return (x@classgamma)},
					"ICLvalue"={return (x@ICLvalue)},
					stop("Invalid slot name.")
			)

		}
)

#' @rdname getter-methods
#' @aliases [,ContinuousOptions-method
#'
setMethod(
		f = "[",
		signature = "ContinuousOptions",
		definition = function(x,i,j,drop) {
			switch(EXPR=i,
					"data"={return (x@data)},
			   	"coclusterdata"={return (x@coclusterdata)},
					"datatype"={return (x@datatype)},
			    "semisupervised"={return (x@semisupervised)},
					"model"={return (x@model)},
					"strategy"={return (x@strategy)},
					"message"={return (x@message)},
					"rowproportions"={return (x@rowproportions)},
					"colproportions"={return (x@colproportions)},
					"rowclass"={return (x@rowclass)},
					"colclass"={return (x@colclass)},
					"likelihood"={return (x@likelihood)},
					"rowposteriorprob"={return (x@rowposteriorprob)},
					"colposteriorprob"={return (x@colposteriorprob)},
					"classmean"={return (x@classmean)},
					"classvariance"={return (x@classvariance)},
					"ICLvalue"={return (x@ICLvalue)},
					stop("Invalid slot name.")
			)

		}
)


#' @rdname getter-methods
#' @aliases [,CategoricalOptions-method
setMethod(
  f = "[",
  signature = "CategoricalOptions",
  definition = function(x,i,j,drop) {
    switch(EXPR=i,
           "data"={return (x@data)},
           "coclusterdata"={return (x@coclusterdata)},
           "datatype"={return (x@datatype)},
           "semisupervised"={return (x@semisupervised)},
           "model"={return (x@model)},
           "strategy"={return (x@strategy)},
           "message"={return (x@message)},
           "rowproportions"={return (x@rowproportions)},
           "colproportions"={return (x@colproportions)},
           "rowclass"={return (x@rowclass)},
           "colclass"={return (x@colclass)},
           "likelihood"={return (x@likelihood)},
           "rowposteriorprob"={return (x@rowposteriorprob)},
           "colposteriorprob"={return (x@colposteriorprob)},
           "classmean"={return (x@classmean)},
           "ICLvalue"={return (x@ICLvalue)},
					 "hyperparam"={return (x@hyperparam)},
					 stop("Invalid slot name.")
    )
  }
)
#' @rdname summary-methods
#' @aliases summary summary,BinaryOptions-method
setMethod(
    f="summary",
    signature = "BinaryOptions",
    definition = function(object,...) {
      if (object@successful) {
        cat("******************************************************************\n")
        cat("Model Family : Bernoulli Latent block model\n")
        cat("Model Name :",object@model)
        if(object@semisupervised)
        { cat("\nCo-Clustering Type : Semi-Supervised")}
        else
        { cat("\nCo-Clustering Type : Unsupervised")}
        cat("\nICL value: ",object@ICLvalue)
        cat("\n\nModel Parameters..")
        cat("\n\nClass Mean:\n")
        print(object@classmean)
        cat("\nClass Dispersion:\n")
        print(object@classdispersion)
        cat("\nRow proportions: ",   object@rowproportions)
        cat("\nColumn proportions: ",object@columnproportions)
        cat("\nPseudo-likelihood: ", object@likelihood)
				cat("\nhyperparam: ",        object@hyperparam)
        cat("\n******************************************************************\n")
      } else {
        cat("Co-Clustering was not successful.\n")
      }
    }
)
#' @rdname summary-methods
#' @aliases summary summary,ContingencyOptions-method
setMethod(
    f="summary",
    signature = "ContingencyOptions",
    definition = function(object,...) {
      if (object@successful) {
        cat("******************************************************************\n")
        cat("Model Family : Poisson Latent block model\n")
        cat("Model Name :",object@model)
        if(object@semisupervised)
          cat("\nCo-Clustering Type : Semi-Supervised")
        else
          cat("\nCo-Clustering Type : Unsupervised")
		cat("\nICL value: ",object@ICLvalue)
		cat("\n\nModel Parameters..")
        cat("\n\nClass Gamma:\n")
        print(object@classgamma)
        cat("\nRow proportions: ",object@rowproportions)
        cat("\nColumn proportions: ",object@columnproportions)
        cat("\nPseudo-likelihood: ",object@likelihood)
        cat("\n******************************************************************\n")
      } else {
        cat("Co-Clustering was not successful.\n")
      }

    }
)

#' @rdname summary-methods
#' @aliases summary summary,ContinuousOptions-method
#'
setMethod(
    f="summary",
    signature = "ContinuousOptions",
    definition = function(object,...) {
      if (object@successful) {
        cat("******************************************************************\n")
        cat("Model Family : Gaussian Latent block model\n")
        cat("Model Name :",object@model)
        if(object@semisupervised)
          cat("\nCo-Clustering Type : Semi-Supervised")
        else
          cat("\nCo-Clustering Type : Unsupervised")
	    cat("\nICL value: ",object@ICLvalue)
	    cat("\n\nModel Parameters..")
        cat("\n\nClass Mean:\n")
        print(object@classmean)
        cat("\nClass Variance:\n")
        print(object@classvariance)
        cat("\nRow proportions: ",object@rowproportions)
        cat("\nColumn proportions: ",object@columnproportions)
        cat("\nPseudo-likelihood: ",object@likelihood)
        cat("\n******************************************************************\n")
      } else {
        cat("Co-Clustering was not successful.\n")
      }

    }
)

#' @rdname summary-methods
#' @aliases summary summary,CategoricalOptions-method
setMethod(
    f="summary",
    signature = "CategoricalOptions",
    definition = function(object,...) {
      if (object@successful) {
        cat("******************************************************************\n")
        cat("Model Family : Categorical Latent block model\n")
        cat("Model Name :",object@model)
        if(object@semisupervised)
          cat("\nCo-Clustering Type : Semi-Supervised")
        else
          cat("\nCo-Clustering Type : Unsupervised")
        cat("\nICL value: ",object@ICLvalue)
        cat("\n\nModel Parameters..")
        cat("\n\nClass Mean:\n")
        cat("******************************************************************\n")
        tempmatrix<-matrix(nrow=object@nbcocluster[1],ncol=object@nbcocluster[2])
        for(h in 1:length(object@classmean)){
          cat("\nProbability for category",h,"in various blocks\n")
          cat("********************************************\n")
          for(k in 1:object@nbcocluster[1]){
            for(l in 1:object@nbcocluster[2]){
              tempmatrix[k,l] = as.double(object@classmean[[h]][[k]][[l]])
            }
          }
          print(tempmatrix)
        }
        cat("\nRow proportions: ",object@rowproportions)
        cat("\nColumn proportions: ",object@columnproportions)
        cat("\nPseudo-likelihood: ",object@likelihood)
				cat("\nhyperparam: ",object@hyperparam)
				cat("\n******************************************************************\n")
      } else {
        cat("Co-Clustering was not successful.\n")
      }
    }
)

#' @rdname plot-methods
#' @aliases plot,BinaryOptions-method
setMethod(
    f="plot",
    signature = c("BinaryOptions"),
    definition = function(x, y , ...)
    {
      if(x@successful)
      {
        data<-x@data
        clusterdata<-x@coclusterdata
        #min of data
        min <- min(data)
        #max of data
        max <- max(data)
        dimention<-dim(data)

        # x11(width=12, height=6)
        layout(matrix(data=c(1,2,3), nrow=1, ncol=3),widths =c(3,6,6))

        #par(mfrow=c(1,2))
        Color<- gray((0:1))

        ColorLevels <- seq(min, max, length=length(Color))
        # Reverse Y axis
        reverse <- nrow(data):1
        data <- data[reverse,]
        clusterdata <- clusterdata[reverse,]

        vararg = list(...)

        coclusterplot = TRUE

        if(hasArg("type"))
        {
          if(vararg$type!="cocluster"&&vararg$type!="distribution")
            stop("Invalid plot type, Valid types are:'cocluster' and 'distribution'")

          if(vararg$type=="distribution")
            coclusterplot = FALSE
        }

        asp = FALSE

        if(hasArg(asp))
        {
          if(vararg$asp==TRUE)
            asp = TRUE
        }

        if(coclusterplot)
        {

          plot(c(0),type="n",xlab="", ylab="", axes=FALSE)
          legend("center",legend =c("0","1"),
              ,border=F,bg = 'gray90', fill=c("black","white"),title="Legend") #pt.bg=c(1,2))
          # Data
          #par(mar = c(1,1,1.5,1))
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])


          title("Original Data")
          # Data cluter
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])

          rowvec=1:(x@nbcocluster[1])
          for (i in 1:x@nbcocluster[1]) {
            rowvec[i]=sum(x@rowclass==i-1)/nrow(x@data)
          }
          rowvec[1]

          colvec=1:(x@nbcocluster[2])
          for (i in 1:x@nbcocluster[2]) {
            colvec[i]=sum(x@colclass==i-1)/ncol(x@data)
          }
          reverse<-x@nbcocluster[1]:1
          abline(h=cumsum(rowvec[reverse])[1:x@nbcocluster[1]-1],v=cumsum(colvec)[1:x@nbcocluster[2]-1], col="blue",lwd=2)

          title("Co-Clustered Data")

        }
        else
        {

          ############### Frequencies of classes of binary data #######################
          par(mfrow=c(x@nbcocluster[1]+1,x@nbcocluster[2]+1))
          clusterdata<-x@coclusterdata
          # Mixture of columns
          cumSample=0
          for (i in 1:x@nbcocluster[1]){
            cumVar=0
            test=0
            for(j in 1:x@nbcocluster[2]){
              tableBloc=table(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))])

              test=rbind(test,tableBloc) #(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]))
              barplot(tableBloc/sum(tableBloc),xlab=paste("Data values block (",i,",",j,")"),ylab="Frequency",main=paste("Block (",i,",",j,")"),col=c("black","white"))
              #factor(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]
              cumVar<-cumVar+sum(x@colclass==j-1)
            }
            barplot(test/sum(test),xlab=paste("Data values of row",i),ylab="frequency", main=paste("Mixture of row",i))
            cumSample<-cumSample+sum(x@rowclass==i-1)
          }

          # Mixture of rows
          cumVar=0
          for (j in 1:x@nbcocluster[2]){
            cumSample=0
            temp=0
            for(i in 1:x@nbcocluster[1]){
              temp=rbind(temp,table(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]))
              cumSample<-cumSample+sum(x@rowclass==i-1)
            }
            barplot(temp/sum(temp),xlab=paste("Data values of column",j),ylab="Frequency", main=paste("Mixture of column",j))
            cumVar<-cumVar+sum(x@colclass==j-1)
          }
          barplot(table(clusterdata)/sum(table(clusterdata)),xlab="Data values",ylab="Frequency",
              main="Final mixture", col=c("black","white"))

          title(main="Histogram/density for each block",outer=TRUE,line=-1)
        }

      }
      else{
        cat("Co-Clustering was not successful.\n")
      }
    }
)

#' @rdname plot-methods
#' @aliases plot,ContingencyOptions-method
setMethod(
    f="plot",
    signature = c("ContingencyOptions"),
    definition = function(x,y,...) {
      if(x@successful)
      {
        data<-x@data
        clusterdata<-x@coclusterdata
        min <- min(data)
        max <- max(data)
        dimention<-dim(data)
        reverse <- nrow(data):1
        data <- data[reverse,]
        clusterdata <- clusterdata[reverse,]
        vararg = list(...)
        coclusterplot = TRUE
        if(hasArg("type"))
        {
          if(vararg$type!="cocluster"&&vararg$type!="distribution")
            stop("Invalid plot type, Valid types are:'cocluster' and 'distribution'")

          if(vararg$type=="distribution")
            coclusterplot = FALSE
        }

        asp = FALSE

        if(hasArg(asp))
        {
          if(vararg$asp==TRUE)
            asp = TRUE
        }

        if(coclusterplot)
        {
          layout(matrix(data=c(1,2,3), nrow=1, ncol=3),widths =c(6,6,1))
          Color<- gray((0:256)/256)
          ColorLevels <- seq(min, max, length=length(Color))

          # Data
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])
          title("Original Data")

          # Data cluster
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])

          rowvec=1:(x@nbcocluster[1])
          for (i in 1:x@nbcocluster[1]) {
            rowvec[i]=sum(x@rowclass==i-1)/nrow(x@data)
          }

          colvec=1:(x@nbcocluster[2])
          for (i in 1:x@nbcocluster[2]) {
            colvec[i]=sum(x@colclass==i-1)/ncol(x@data)
          }
          reverse<-x@nbcocluster[1]:1
          abline(h=cumsum(rowvec[reverse])[1:x@nbcocluster[1]-1],v=cumsum(colvec)[1:x@nbcocluster[2]-1],col="blue",lwd=2)
          title("Co-Clustered Data")

          par(mar = c(1,2,1.5,1))
          image(1, ColorLevels,
              matrix(data=ColorLevels, ncol=length(ColorLevels),nrow=1),
              col=Color,xaxt="n")
          title("Scale")
        }
        else
        {
          ###########  barplot of contingency data  ##########################

          par(mfrow=c(x@nbcocluster[1]+1,x@nbcocluster[2]+1))
          clusterdata<-x@coclusterdata
          cumSample=0
          for (i in 1:x@nbcocluster[1]){

            cumVar=0
            for(j in 1:x@nbcocluster[2])
            {
              blocTable=clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]
              barplot(table(blocTable)/sum(table(blocTable)), main=paste("Block (",i,",",j,")"),
                  xlab=paste("Data values of block (",i,",",j,")"), col="lightblue",ylab="Frequency")
              cumVar<-cumVar+sum(x@colclass==j-1)
            }
            rowBloc<-clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)), 1:(dim(clusterdata)[2])]
            barplot(table(rowBloc)/sum(table(rowBloc)),col="lightblue",xlab=paste("Data values of row",i),ylab="Frequency")

            cumSample<-cumSample+sum(x@rowclass==i-1)
          }

          cumVar=0

          for (k in 1:x@nbcocluster[2])
          {

            colBloc=table(clusterdata[1:dim(clusterdata)[1],(cumVar+1):(cumVar+sum(x@colclass==k-1))])
            barplot(colBloc/sum(colBloc),ylab="Frequency",xlab=paste("Data values of column",k),
                col="lightblue", main=paste("Mixture density of column",k))

            cumVar<-cumVar+sum(x@colclass==k-1)
          }
          barplot(table(clusterdata)/sum(table(clusterdata)), xlab=paste("Data values"),ylab="Frequency",
              col="lightblue",main="Final mixture density")

          #title(main="Histograms of classes of contingency data",outer=TRUE,line=-1)
        }
      }
      else{
        cat("Co-Clustering was not successful.\n")
      }
    }
)

#' @rdname plot-methods
#' @aliases plot,ContinuousOptions-method
setMethod(
    f="plot",
    signature = c("ContinuousOptions"),
    definition = function(x,y,...) {
      if(x@successful)
      {
        vararg = list(...)
        coclusterplot = TRUE
        if(hasArg("type"))
        {
          if(vararg$type!="cocluster"&&vararg$type!="distribution")
            stop("Invalid plot type, Valid types are:'cocluster' and 'distribution'")

          if(vararg$type=="distribution")
            coclusterplot = FALSE
        }
        asp = FALSE
        if(hasArg(asp))
        {
          if(vararg$asp==TRUE)
            asp = TRUE
        }

        data<-x@data
        clusterdata<-x@coclusterdata

        min <- min(data)
        max <- max(data)
        dimention<-dim(data)

        reverse <- nrow(data):1
        data <- data[reverse,]
        clusterdata <- clusterdata[reverse,]

        if (coclusterplot)
        {
          layout(matrix(data=c(1,2,3), nrow=1, ncol=3),widths =c(6,6,1))
          Color<- gray((0:256)/256)
          ColorLevels <- seq(min, max, length=length(Color))
          # Data
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])
          title("Original Data")

          # Data cluter
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])

          rowvec=1:(x@nbcocluster[1])
          for (i in 1:x@nbcocluster[1]) {
            rowvec[i]=sum(x@rowclass==i-1)/nrow(x@data)
          }

          colvec=1:(x@nbcocluster[2])
          for (i in 1:x@nbcocluster[2]) {
            colvec[i]=sum(x@colclass==i-1)/ncol(x@data)
          }
          reverse<-x@nbcocluster[1]:1
          abline(h=cumsum(rowvec[reverse])[1:x@nbcocluster[1]-1],v=cumsum(colvec)[1:x@nbcocluster[2]-1],col="blue",lwd=2)

          title("Co-Clustered Data")

          par(mar = c(1,2,1.5,1))
          image(1, ColorLevels,
              matrix(data=ColorLevels, ncol=length(ColorLevels),nrow=1),
              col=Color,xaxt="n")
          title("Scale")
        }
        else
        {
          ################ The graph of each block and the mixture ##########################
          # get dimension
          nbRowCluster=x@nbcocluster[1];
          nbColCluster=x@nbcocluster[2];
          # compute the number of varaibles in each  block
          nbVar=vector(length=nbColCluster);
          for(j in 1:nbColCluster)
          {
            nbVar[j] = sum(x@colclass==j-1);
          }
          # compute the number of sample in each  block
          nbSample=vector(length=nbRowCluster);
          for (i in 1:nbRowCluster)
          {
            nbSample[i] = sum(x@rowclass==i-1);
          }
          # seprate the graphic device
          par(mfrow=c(nbRowCluster+1,nbColCluster+1))
          cumSample=0
          clusterdata<-x@coclusterdata
          # main loop
          colorBloc=1
          rowcolorMixt=1

          for (i in 1:nbRowCluster)
          {
            densMixt=0
            cumVar=0
            mindata = min(clusterdata[(cumSample+1):(cumSample+nbSample[i]),1:(dim(clusterdata)[2])])
            maxdata = max(clusterdata[(cumSample+1):(cumSample+nbSample[i]),1:(dim(clusterdata)[2])])
            xaxis =seq( mindata, maxdata , by=1)
            densMixt=vector(length = length(xaxis))

            for(j in 1:nbColCluster)
            {
              tableBloc<-clusterdata[(cumSample+1):(cumSample+nbSample[i]), (cumVar+1):(cumVar+nbVar[j])]
              hist(tableBloc,xlab=paste("Data values of block (",i,",",j,")"),
                  col="lightblue",ylab="Frequency", main=paste("Block (",i,",",j,")"),proba=TRUE)

              densBloc <- dnorm( xaxis,x@classmean[i,j],sqrt(x@classvariance[i,j]))
              densMixt = densMixt + x@columnproportions[j]*densBloc
              points(x=xaxis, y=densBloc,type='l',lty=1, col=colorBloc+1)
              cumVar<-cumVar+nbVar[j]
              colorBloc=colorBloc+1
            }

            # Graph of the mixture of columns
            colBloc<-clusterdata[(cumSample+1):(cumSample+nbSample[i]), 1:(dim(clusterdata)[2])]
            h<-hist(colBloc,col="lightblue",xlab=paste("Data values of row",i),ylab="Frequency", main=paste("Mixture density of row",i),proba=TRUE)
            points(xaxis,densMixt,col="black",type='l',lty=1)

            for(l in 1:nbColCluster)
            {
              densBloc<-x@columnproportions[l]*dnorm( xaxis,x@classmean[i,l],sqrt(x@classvariance[i,l]))
              points( xaxis,densBloc,type='l',lty=1,col=rowcolorMixt+1)
              rowcolorMixt=rowcolorMixt+1
            }
            cumSample<-cumSample+nbSample[i]
          }

          ### bloc columns
          # Graph of the mixture of rows
          cumVar=0
          colColor=2

          for(j in 1:nbColCluster)
          {
            densMixt=0
            rowTable<-clusterdata[1:(dim(clusterdata)[1]),(cumVar+1):(cumVar+nbVar[j])]
            h<-hist(rowTable, xlab=paste("Data values of column",j),ylab="Frequency",
                col="lightblue", main=paste("Mixture density of column",j),proba=TRUE)

            xaxis=seq(min(rowTable), max(rowTable), length.out=1000)
            cl=colColor
            for(i in 1:nbRowCluster)
            {
              densBloc<-x@rowproportions[i]*dnorm( xaxis,x@classmean[i,j],sqrt(x@classvariance[i,j]))
              densMixt=densMixt+ densBloc
              points( xaxis,densBloc,type='l',lty=1,col=cl)
              cl=nbColCluster+colColor
            }
            colColor=colColor+1
            par(new=TRUE, ann= FALSE)
            points( xaxis,densMixt,col="black",type='l',lty=1)
            cumVar<-cumVar+nbVar[j]
          }

          # All mixtures of rows and columns
          hist(clusterdata,xlab=paste("Data values"),ylab="Frequency",
              col="lightblue",main="Final mixture density",proba=TRUE)

          xaxis=seq(min(clusterdata),max(clusterdata), length.out = 1000)
          densMixt=0
          colorMixt=1
          for(i in 1:nbRowCluster){
            for(j in 1:nbColCluster){
              densBloc<-x@rowproportions[i]*x@columnproportions[j]*dnorm(xaxis,x@classmean[i,j],sqrt(x@classvariance[i,j]))
              densMixt<-densMixt+densBloc
              points(xaxis,densBloc, col=colorMixt+1,type='l',lty=1)
              colorMixt=colorMixt+1
            }
          }
          par(new=TRUE, ann= FALSE)
          points(xaxis,densMixt,col="black",type='l',lty=1)#,yaxt ="n",xaxt="n", xlab="",ylab="", main="")

          title(main="Histograms and densities of each block",outer=TRUE,line=-1)
        }

      }
      else{
        cat("Co-Clustering was not successful.\n")
      }
    }
)

#' @rdname plot-methods
#' @aliases plot,CategoricalOptions-method
setMethod(
    f="plot",
    signature = c("CategoricalOptions"),
    definition = function(x,y,...) {
      if(x@successful)
      {
        data<-x@data
        clusterdata<-x@coclusterdata
        min <- min(data)
        max <- max(data)
        dimention<-dim(data)

        reverse <- nrow(data):1
        data <- data[reverse,]
        clusterdata <- clusterdata[reverse,]

        vararg = list(...)

        coclusterplot = TRUE
        if(hasArg("type"))
        {
          if(vararg$type!="cocluster"&&vararg$type!="distribution")
            stop("Invalid plot type, Valid types are:'cocluster' and 'distribution'")

          if(vararg$type=="distribution")
            coclusterplot = FALSE
        }

        asp = FALSE

        if(hasArg(asp))
        {
          if(vararg$asp==TRUE)
            asp = TRUE
        }

        if(coclusterplot)
        {
          layout(matrix(data=c(1,2,3), nrow=1, ncol=3),widths =c(6,6,1))

          Color<- gray((0:256)/256)

          ColorLevels <- seq(min, max, length=length(Color))

          # Data
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(data), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])
          title("Original Data")


          # Data cluter
          par(mar = c(1,1,1,1))
          if(asp == FALSE)
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max))
          else
            image(t(clusterdata), col=Color, axes=FALSE, zlim=c(min,max), asp=dim(data)[1]/dim(data)[2])


          rowvec=1:(x@nbcocluster[1])
          for (i in 1:x@nbcocluster[1]) {
            rowvec[i]=sum(x@rowclass==i-1)/nrow(x@data)
          }

          colvec=1:(x@nbcocluster[2])
          for (i in 1:x@nbcocluster[2]) {
            colvec[i]=sum(x@colclass==i-1)/ncol(x@data)
          }
          reverse<-x@nbcocluster[1]:1
          abline(h=cumsum(rowvec[reverse])[1:x@nbcocluster[1]-1],v=cumsum(colvec)[1:x@nbcocluster[2]-1],col="blue",lwd=2)
          title("Co-Clustered Data")

          par(mar = c(1,2,1.5,1))
          image(1, ColorLevels,
              matrix(data=ColorLevels, ncol=length(ColorLevels),nrow=1),
              col=Color,xaxt="n")
          title("Scale")
        }
        else
        {
          ############### Frequencies of classes of categorical data #######################
          par(mfrow=c(x@nbcocluster[1]+1,x@nbcocluster[2]+1))
          clusterdata<-x@coclusterdata
          # Mixture of columns
          cumSample=0
          for (i in 1:x@nbcocluster[1]){
            cumVar=0
            test=0
            for(j in 1:x@nbcocluster[2]){
              tableBloc=table(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))])

              test=rbind(test,tableBloc) #(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]))
              barplot(tableBloc/sum(tableBloc),xlab=paste("Data values block (",i,",",j,")"),ylab="Frequency",main=paste("Block (",i,",",j,")"))
              #factor(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]
              cumVar<-cumVar+sum(x@colclass==j-1)
            }
            barplot(test/sum(test),xlab=paste("Data values of row",i),ylab="frequency", main=paste("Mixture of row",i))
            cumSample<-cumSample+sum(x@rowclass==i-1)
          }

          # Mixture of rows
          cumVar=0
          for (j in 1:x@nbcocluster[2]){
            cumSample=0
            temp=0
            for(i in 1:x@nbcocluster[1]){
              temp=rbind(temp,table(clusterdata[(cumSample+1):(cumSample+sum(x@rowclass==i-1)),(cumVar+1):(cumVar+sum(x@colclass==j-1))]))
              cumSample<-cumSample+sum(x@rowclass==i-1)
            }
            barplot(temp/sum(temp),xlab=paste("Data values of column",j),ylab="Frequency", main=paste("Mixture of column",j))
            cumVar<-cumVar+sum(x@colclass==j-1)
          }
          barplot(table(clusterdata)/sum(table(clusterdata)),xlab="Data values",ylab="Frequency",
              main="Final mixture")

          title(main="Histogram/density for each block",outer=TRUE,line=-1)
        }

      }
      else{
        cat("Co-Clustering was not successful.\n")
      }

    }
)