GGenemy: Visualization of multidimensional data by multiple conditioning

Documented in plot_GGenemy

#' Plot GGenemy
#'
#' Plots densities of continuous variables, conditioned on quantiles of
#' continuous variables or categories of categorical variables. For categorical
#' variables, boxplots or barplots will be created, depending on whether the
#' variable to be conditioned on is continuous or categorical.
#' Either the user can use a number of same sized quantiles or
#' can select the size of own quantiles
#'
#'
#' @param dataset A dataframe.
#' @param given_var A variable from your \code{dataset}.
#' @param var_to_plot The variables to plot. When not selecting any variable, all variables from your \code{dataset}
#'   will be plotted.
#' @param n_quantiles Number of quantiles you want to partition \code{given_var}
#'   into.
#' @param boxplot If TRUE, all numerical variables will be presented as boxplots 
#' instead of densities.
#' You can also provide a vector with the names of the numerics to show
#' as boxplots if you want to use this option only for a selection of variables.
#' @param selfrange Vector of n*2 values or matrix with 2 columns.
#' The first value as min and the second value as max of the self selected quantile.
#' @param remaining logical. If TRUE, the remaining values not within selfrange will be plotted to
#' category named 'remaining'.
#'
#'
#'
#' @return Several ggplot graphics. The total number of plots will equal the
#'   number of variables in your \code{dataset} or the amount of variables 
#'   selected in \code{var_to_plot}.
#' @export
#'
#' @examples
#' n <- 500
#' number_variables <- 5
#'
#' set.seed(42)
#' variables <- as.data.frame(sapply(1:number_variables, function(a) {
#'   if (runif(1) < 0.5) {
#'     rnorm(n)
#'   } else {
#'     rgamma(n, 0.5)
#'   }
#' }))
#'
#'
#' plot_GGenemy(variables, "V1")
#'
plot_GGenemy <- function(dataset, given_var, var_to_plot = NULL, n_quantiles = 5,
                         boxplot = FALSE, selfrange = NULL, remaining = TRUE) {
  if (length(class(dataset)) > 1) {
    dataset <- unclass(dataset)
    dataset <- as.data.frame(dataset)
  }

  if (!is.null(selfrange)) {
    if (is.factor(dataset[, given_var])) {
      data_help <- dataset
      data_help$quant <- "remaining"
      for (i in length(selfrange):1) {
        quanthelp <- which(dataset[, given_var] == selfrange[i])
        data_help$quant[quanthelp] <- paste(selfrange[i])
      }
      if (remaining == FALSE) {
        data_help <- data_help[-which(data_help$quant == "remaining"), ]
      }
    } else if (is.matrix(selfrange) | length(selfrange) %% 2 == 0) {
      if (is.matrix(selfrange)) {
        matrixquant <- selfrange
      } else {
        matrixquant <- matrix(selfrange, ncol = 2, byrow = TRUE)
      }
      if(any(is.na(matrixquant))){
        which_na <- which(is.na(t(matrixquant)))
        hel <- c()
        for(i in 1:length(which_na)){
          if(which_na[i] %% 2 == 0){
            hel[c(1,2)+(i-1)*2] <- c(which_na[i],which_na[i]-1)  
          } else {
            hel[c(1,2)+(i-1)*2] <- c(which_na[i],which_na[i]+1)  
          }
        }
        uniq <- unique(hel)
        matrixquant <- matrix(t(matrixquant)[-uniq],byrow = TRUE, ncol = 2)
      }
      data_help <- dataset
      data_help$quant <- 999
      labs <- NULL
      for (i in nrow(matrixquant):1) {
        quanthelp <- which(dataset[, given_var] >= matrixquant[i, 1] & dataset[, given_var] <= matrixquant[i, 2])
        helper <- duplicated(as.vector(t(matrixquant)))[c(-1, 0) + i * 2]
        if (any(helper)) {
          if (helper[1] & !all(helper)) {
            data_help$quant[quanthelp] <- i
            labs[i] <- paste(">", matrixquant[i, 1], "to", "<=", matrixquant[i, 2])
          } else if (helper[2] & !all(helper)) {
            data_help$quant[quanthelp] <- i
            labs[i] <- paste(">=", matrixquant[i, 1], "to", "<", matrixquant[i, 2])
          } else {
            data_help$quant[quanthelp] <- i
            labs[i] <- paste(">", matrixquant[i, 1], "to", "<", matrixquant[i, 2])
          }
        } else {
          data_help$quant[quanthelp] <- i
          labs[i] <- paste(">=", matrixquant[i, 1], "to", "<=", matrixquant[i, 2])
        }
      }
      

      # if (remaining == FALSE) {
      #   data_help <- data_help[-which(data_help$quant == "remaining"), ]
      # }
    } else {
      stop("For selfrange: Insert a Vector of quantiles or a matrix with the quantiles ordered by row")
    }
    
    if (remaining == TRUE) {
      labs <- c(labs, "remaining")
    }

    if (remaining == FALSE) {
      data_help <- data_help[-which(data_help$quant == 999), ]
    }
    data_help$quant <- factor(data_help$quant, labels = labs)
    
   } else {
    if (is.factor(dataset[, given_var])) {
      if (n_quantiles != length(levels(dataset[, given_var]))) {
        message(paste0(given_var, " is a categorical variable. The number of categories will be defined as a condition."))
      }
      data_help <- dataset
      data_help$quant <- dataset[, given_var]
    } else {
      var_goal <- dplyr::select(dataset, given_var)[, 1]
      quantiles <- stats::quantile(var_goal, 1:(n_quantiles - 1) / (n_quantiles))
      quantiles <- as.numeric(quantiles)
      data_help <- dataset
      data_help$quant <- 1 + findInterval(var_goal, quantiles)
      data_help$quant <- as.factor(data_help$quant)
      minim <- round(min(var_goal), 3)
      maxim <- round(max(var_goal), 3)
      quantiles_round <- round(quantiles, 3)
      maxposs <- 999999
      minposs <- 0.0001
      if (quantiles[1] < minposs | quantiles[length(quantiles)] > maxposs) {
        for (i in 1:length(quantiles)) {
          if (quantiles[i] < minposs | quantiles[i] > maxposs) {
            quantiles_round[i] <- formatC(quantiles[i], format = "e", digits = 2)
          }
        }
      }
      if (min(var_goal) < minposs | max(var_goal) > maxposs) {
        if (min(var_goal) < minposs){
        minim <- formatC(min(var_goal), format = "e", digits = 2)
        }
        if (max(var_goal) > maxposs) {
        maxim <- formatC(max(var_goal), format = "e", digits = 2)
        }
      }
      for (i in 1:n_quantiles) {
        if (i == 1) {
          levels(data_help$quant)[i] <- paste(minim, "to", quantiles_round[i])
        } else if (!(i == length(levels(data_help$quant)))) {
          levels(data_help$quant)[i] <- paste(quantiles_round[i - 1], "to", quantiles_round[i])
        } else {
          levels(data_help$quant)[i] <- paste(quantiles_round[i - 1], "to", maxim)
        }
      }
    }
  }


  g <- ggplot2::ggplot(data_help, ggplot2::aes(fill = quant)) + ggplot2::theme_minimal() +
    ggplot2::theme(plot.title = ggplot2::element_text(size = 15, face = "bold"))

  plotit <- function(a) {
    if (is.factor(data_help[, a]) & is.factor(data_help[, given_var])) {
      g + ggplot2::geom_bar(ggplot2::aes_string(x = a, fill = given_var)) +
        ggplot2::ggtitle(paste0(a, " conditional on ", given_var))
    } else if (is.factor(data_help[, a])) {
      g + ggplot2::geom_boxplot(ggplot2::aes_string(x = a, y = given_var)) +
        ggplot2::ggtitle(paste0(a, " conditional on ", given_var))
    } else if (!is.logical(boxplot) & any(boxplot == a)) {
      g + ggplot2::geom_boxplot(ggplot2::aes_string(x = given_var, y = a)) +
        ggplot2::ggtitle(paste0(a, " conditional on ", given_var))
    } else if (any(boxplot == TRUE)) {
      g + ggplot2::geom_boxplot(ggplot2::aes_string(x = given_var, y = a)) +
        ggplot2::ggtitle(paste0(a, " conditional on ", given_var))
    } else {
      g + ggplot2::geom_density(alpha = 0.7) + ggplot2::aes_string(x = a) +
        ggplot2::ggtitle(paste0(a, " conditional on ", given_var))
    }
  }

  if (all(c(!is.character(var_to_plot), !is.null(var_to_plot)))) {
    stop(paste0(var_to_plot, " has to be a character or leave it as NULL to calculate
                   conditional densities for all other variables of the dataset."))
  } else if (is.null(var_to_plot)) {
    names_var <- names(dataset)
    lapply(names_var, plotit)
  } else if (length(var_to_plot) == 1) {
    plotit(var_to_plot)
  } else {
    lapply(var_to_plot, plotit)
  }
}