pkg/R/plot_valse.R

   1 #' Plot
   2 #'
   3 #' It is a function which plots relevant parameters
   4 #'
   5 #' @param X matrix of covariates (of size n*p)
   6 #' @param Y matrix of responses (of size n*m)
   7 #' @param model the model constructed by valse procedure
   8 #' @param comp TRUE to enable pairwise clusters comparison
   9 #' @param k1 index of the first cluster to be compared
  10 #' @param k2 index of the second cluster to be compared
  11 #'
  12 #' @importFrom ggplot2 ggplot aes ggtitle geom_tile geom_line geom_point scale_fill_gradient2 geom_boxplot theme
  13 #' @importFrom cowplot background_grid
  14 #' @importFrom reshape2 melt
  15 #'
  16 #' @export
  17 plot_valse <- function(X, Y, model, comp = FALSE, k1 = NA, k2 = NA)
  18 {
  19   n <- nrow(X)
  20   K <- length(model$pi)
  21   ## regression matrices
  22   gReg <- list()
  23   for (r in 1:K)
  24   {
  25     Melt <- melt(t((model$phi[, , r])))
  26     gReg[[r]] <- ggplot(data = Melt, aes(x = Var1, y = Var2, fill = value))  +
  27       geom_tile() + scale_fill_gradient2(low = "blue", high = "red", mid = "white",
  28       midpoint = 0, space = "Lab") + ggtitle(paste("Regression matrices in cluster", r))
  29   }
  30   print(gReg)
  31
  32   ## Differences between two clusters
  33   if (comp)
  34   {
  35     if (is.na(k1) || is.na(k2))
  36       print("k1 and k2 must be integers, representing the clusters you want to compare")
  37     Melt <- melt(t(model$phi[, , k1] - model$phi[, , k2]))
  38     gDiff <- ggplot(data = Melt, aes(x = Var1, y = Var2, fill = value)) +
  39       geom_tile() + scale_fill_gradient2(low = "blue", high = "red", mid = "white", midpoint = 0,
  40         space = "Lab") + ggtitle(paste("Difference between regression matrices in cluster",
  41         k1, "and", k2))
  42     print(gDiff)
  43   }
  44
  45   ### Covariance matrices
  46   matCov <- matrix(NA, nrow = dim(model$rho[, , 1])[1], ncol = K)
  47   for (r in 1:K)
  48     matCov[, r] <- diag(model$rho[, , r])
  49   MeltCov <- melt(matCov)
  50   gCov <- ggplot(data = MeltCov, aes(x = Var1, y = Var2, fill = value)) + geom_tile() +
  51     scale_fill_gradient2(low = "blue", high = "red", mid = "white", midpoint = 0,
  52       space = "Lab") + ggtitle("Covariance matrices (diag., one row per cluster)")
  53   print(gCov)
  54
  55   ### Proportions
  56   gam2 <- matrix(NA, ncol = K, nrow = n)
  57   for (i in 1:n)
  58     gam2[i, ] <- c(model$proba[i, model$affec[i]], model$affec[i])
  59
  60   bp <- ggplot(data.frame(gam2), aes(x = X2, y = X1, color = X2, group = X2)) + geom_boxplot() +
  61      theme(legend.position = "none") + background_grid(major = "xy", minor = "none")  +
  62     ggtitle("Assignment boxplot per cluster")
  63   print(bp)
  64 }