R/plot.R

In dynsurv: Dynamic Models for Survival Data

##
## R package dynsurv by Wenjie Wang, Ming-Hui Chen, Xiaojing Wang, and Jun Yan
## Copyright (C) 2011-2023
##
## This file is part of the R package dynsurv.
##
## The R package dynsurv is free software: You can redistribute it and/or
## modify it under the terms of the GNU General Public License as published by
## the Free Software Foundation, either version 3 of the License, or any later
## version (at your option). See the GNU General Public License at
## <https://www.gnu.org/licenses/> for details.
##
## The R package dynsurv is distributed in the hope that it will be useful,
## but WITHOUT ANY WARRANTY without even the implied warranty of
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
##


##' Plot Coefficient Function
##'
##' Plot coefficient values formatted in a data frame returned by function
##' \code{coef}.
##'
##' To plot estimated coefficient functions from different models together, one
##' can first combine the data frames returned by \code{coef}, and then call
##' \code{plotCoef}, for example, \code{plotCoef(rbind(coef(fit1),
##' coef(fit2)))}.
##'
##' To specify the time range of the plot, one can either utilize the
##' \code{ggplot} functionality, say \code{plotCoef(coef(fit)) + xlim(2, 10)};
##' or manipulate the data frame first before calling \code{plotCoef}, e.g.,
##' \code{plotCoef(subset(coef(fit), Time > 2 & Time < 10))}.
##'
##' @usage plotCoef(object, smooth = FALSE, ...)
##' @param object A data.frame returned by function \code{coef}.
##' @param smooth A logical value, default \code{FALSE}. If \code{TRUE}, plot
##' the coefficients as smooth lines; otherwise, plot the coefficients as
##' piece-wise constant step functions.
##' @param ... Other arguments.
##' @return A \code{ggplot} object.
##' @seealso \code{\link{coef.bayesCox}}, \code{\link{coef.splineCox}}, and
##' \code{\link{coef.tvTran}}.
##' @keywords plot coefficient
##' @examples
##' ## See the examples in bayesCox, splineCox, and tvTran.
##' @importFrom ggplot2 ggplot aes_string geom_step geom_line facet_wrap
##' facet_grid ylab theme
##' @importFrom grid unit
##' @export
plotCoef <- function(object, smooth = FALSE, ...)
{
    p <- ggplot(data = object, aes_string(x = "Time"))

    if (!smooth)
        p <- p + geom_step(aes_string(y = "Mid"), direction = "vh") +
            geom_step(aes_string(y = "High"), direction = "vh", linetype = 2) +
            geom_step(aes_string(y = "Low"), direction = "vh", linetype = 2)
    else
        p <- p + geom_line(aes_string(y = "Mid")) +
            geom_line(aes_string(y = "High"), linetype = 2) +
            geom_line(aes_string(y = "Low"), linetype = 2)

    if (length(unique(object$Model)) == 1)
        p <- p + facet_wrap(~ Cov, scales = "free_y")
    else
        p <- p + facet_grid(Cov ~ Model, scales = "free_y")

    p <- p + ylab("Coefficient") +
        theme(plot.margin = unit(rep(0, 4), "lines"))

    p
}


##' Plot Jump Information in Bayesian Dynamic Model
##'
##' \code{plotJumpTrace} plots the MCMC history of the number of pieces.
##' \code{plotJumpHist} plots the histogram of the number of pieces. The input
##' data frame is returned by function \code{jump}.
##'
##' @param object A data.frame returned by function \code{jump}.
##' @param ... Other arguments.
##' @return A \code{ggplot} object.
##' @seealso \code{\link{jump.bayesCox}}.
##' @keywords plot jump
##' @examples
##' ## See the examples in bayesCox
##' @name plotJump
NULL


##' @rdname plotJump
##' @aliases plotJumpTrace
##' @usage plotJumpTrace(object, ...)
##' @importFrom ggplot2 ggplot aes_string geom_line facet_wrap xlab ylab theme
##' @importFrom grid unit
##' @export
plotJumpTrace <- function(object, ...)
{
    ggplot(data = object, aes_string(x = "Iter", y = "Count")) +
        geom_line(size = 0.1, alpha = 0.6) +
        facet_wrap(~ Cov) +
        xlab("Iteration") + ylab("Pieces of Coefficient") +
        theme(plot.margin = unit(rep(0, 4), "lines"))
}


##' @rdname plotJump
##' @aliases plotJumpHist
##' @usage plotJumpHist(object, ...)
##' @importFrom ggplot2 ggplot aes_string stat_bin facet_wrap xlab ylab theme
##' @importFrom grid unit
##' @export
plotJumpHist <- function(object, ...)
{
    ## p <- ggplot(data = object, aes(x = factor(Count))) +
    ggplot(data = object, aes_string(x = "Count")) +
        ## stat_bin(aes(y  = ..count../sum(..count..))) +
        stat_bin(aes_string(y  =  "..density..")) +
        facet_wrap(~ Cov) +
        xlab("Pieces of Coefficient") +
        ylab("Relative Frequency") +
        theme(plot.margin = unit(rep(0, 4), "lines"))
}


##' Plot Latent Variance in Bayesian Cox Model
##'
##' Plot the latent variance \code{nu} when the hierarchical AR(1) process
##' prior is used for the \code{bayesCox} model. It is applicable when
##' \code{model="TimeVarying"} or \code{model="Dynamic"}, and
##' \code{coef.prior=list(type="HAR1")}. The input data frame is returned by
##' function \code{nu}.
##'
##' @usage plotNu(object, ...)
##' @aliases plotNu
##' @param object A data.frame returned by the function \code{nu}.
##' @param ... Other arguments.
##' @return A \code{ggplot} object.
##' @seealso \code{\link{nu.bayesCox}}.
##' @keywords plot latent variance
##' @examples
##' ## See the examples in bayesCox
##' @importFrom ggplot2 ggplot aes_string stat_bin xlab ylab theme facet_wrap
##' facet_grid
##' @importFrom grid unit
##' @export
plotNu <- function(object, ...)
{
    cnt <- "..density.."
    ## p <- ggplot(data = object, aes_string(x = "Value")) +
    p <- ggplot(data = object, aes_string(x = "Value")) +
        ##  stat_bin(aes(y  = ..count../sum(..count..))) +
        stat_bin(aes_string(y  =  cnt)) +
        xlab("Nu") + ylab("Relative Frequency") +
        theme(plot.margin = unit(rep(0, 4), "lines"))

    if (length(levels(factor(object$Model))) == 1)
        p <- p + facet_wrap(~ Cov)
    else
        p <- p + facet_grid(Cov ~ Model)
    p
}