R/plots.R

In jcizel/SovereignCrisis: What the package does (one line)

.tsLinePlot <- function(data,
                        x,
                        y,
                        limits = as.Date(c('1990-01-01','2013-01-01')),
                        format = "%Y",
                        xlabel = "",
                        ylabel = ""){
    ## require(scales)
    o <- ggplot(data = data)
    o <- o + geom_step(aes_string(x = x,
                                  y = y))
    o <- o + theme_bw()
    o <- o + theme(axis.text.x = element_text(angle = 90, hjust = 1),
                   axis.text.y = element_text(size = 10))
    o <- o +  scale_x_date(labels = date_format(format = format),
                           limits = limits)
    o <- o + labs(x = xlabel,
                  y = ylabel)
    return(o)
}

.crisisRegion <- function(plot,
                          crisisdata){
    if (nrow(crisisdata)==0){
        o <- plot
    } else {
        o <- plot + geom_rect(data = crisisdata,
                              aes(xmin = mind.event,
                                  xmax = maxd.event,
                                  ymin=-Inf, ymax=Inf,
                                  fill = t),
                              fill="red",
                              alpha = 0.2,
                              inherit.aes = FALSE)
    }
    return(o)
}

## x = "ARG"
## o <- .tsLinePlot(data = getSPRatings()[iso3==x],
##                  x='date',
##                  y='ratingnum',
##                  limits = as.Date(c('1990-01-01','2013-01-01'))
##                  ) + .crisisRegion(data = crises2[iso3==x])

## filename = '~/Downloads/test.pdf'
## ggsave(plot = o,
##        filename = filename)


.removeSpecialCharacters <- function(str){
    o <- gsub(pattern = "[[:punct:]]",".",str)
    return(o)
}

##' .. content for \description{} (no empty lines) ..
##'
##' .. content for \details{} ..
##' @title Line plots of sovereign credit ratings, CDS spreads, and government
##' bond spreads 
##' @param isoSel ISO3 country code of a country for which the plot is to be created. 
##' @param limits a Date vector indicating the time range of the plot
##' @param filename path of the resulting plot (in pdf format)
##' @return NULL
##' @author Janko Cizel
plotSovBenchmarks <- function(isoSel = "ARG",
                              crisisdb = loadCrisisDB(),
                              crisistype = 'debtcrisis',
                              limits = as.Date(c('1990-01-01','2013-01-01')),
                              filename = '~/Downloads/test.pdf',
                              width = 320,
                              height = 420,
                              plotDefinition =
                                  list('ratingnum' =
                                           list(data = getSPRatings(),
                                                y = 'ratingnum',
                                                ylabel = 'S&P Sovereign Credit Rating',
                                                idCol = 'iso3'),
                                       'cds' =
                                           list(data = getBloombergSovCDS(),
                                                y = 'cds',
                                                ylabel = '5-Year Sovereign CDS Spread',
                                                idCol = 'iso3'),
                                       'spread' =
                                           list(data = getSovBondSpreads(),
                                                y = 'spread',
                                                ylabel = 'Sovereign Bond Yield Spread',
                                                idCol = 'iso3')))
{
    crises = crisisdb
    ## crises[, debtcrisis := (`Foreign Sov Debt`*1 + `Domestic Sov Debt`*1)>0]
    crises <- 
        createCrisisVariables(
            crisisDT = crises,
            crisisCol = crisistype,
            idCol = 'iso3',
            timeCol = 'date'
        )
    
    crises2 <- 
        crises[CN>0, {
            mind.event = min(date)
            maxd.event = max(date)
            if (mind.event == maxd.event)
                maxd.event = maxd.event + 365
                
            list(mind.event = mind.event,
                 maxd.event = maxd.event)
        }
             , by = c('iso3', 'CN')]

    plot_list <- list()

    for (x in 1:length(plotDefinition)){
        plot_list[[x]] <- try({
            .t <- plotDefinition[[x]]

            d <- .t$data[get(.t$idCol) == isoSel][, c('date',.t$y), with = FALSE]
            setnames(d,names(d),.removeSpecialCharacters(names(d)))
            
            if (nrow(d)==0) stop("No data available")
            
            o <- .tsLinePlot(data = d,
                             x='date',
                             y=.removeSpecialCharacters(.t$y),
                             limits = limits,
                             ylabel = .t$ylabel
                             ) 
            .crisisRegion(plot = o,
                          crisisdata = crises2[iso3==isoSel])
        })        
    }

    plot_list <- Filter(function(x) !"try-error" %in% class(x), plot_list)
    
    g <- 
        do.call("arrangeGrob",
                c(plot_list,
                  list(
                      as.table = FALSE,
                      ncol = 1,
                      default.units = "cm"
                      )
                  ))

    ggsave(filename = filename,
           width = width,
           height = height,
           units = "mm",
           plot = g)

    return(NULL)
}

## plotSovBenchmarks(isoSel = "ARG")

## plotSovBenchmarks(isoSel = "ARG")
## plotSovBenchmarks(isoSel = "BRA")
## plotSovBenchmarks(isoSel = "ESP")
## plotSovBenchmarks(isoSel = "GRC")
## plotSovBenchmarks(isoSel = "PRT")

.createNoteOnCrisisEpisodes <- function(data){
    data[COUNTDOWN == 0,{
        cat(.BY[[1]],"\n")
        paste(unique(paste0(iso3,"-",year)),
              collapse = "; ")
    }, by = CRISIS][[2]]
}

.densityPlot <- function(data,
                         x = 'ratingnum',
                         xlabel = "",
                         ylabel = "",
                         grouplabel = "Crisis Indicator"){
    episodes <- .createNoteOnCrisisEpisodes(data = data[!is.na(get(x))])
    o <- ggplot(data = data)  
    o <- o + geom_density(aes_string(x = x,
                                     linetype = "as.factor(CRISIS)"
                                     )) 
    o <- o + ggthemes::scale_linetype_stata() 
    o <- o + theme_bw()             
    o <- o + guides(size = FALSE) 
    o <- o + theme(axis.text.x = element_text(angle = 90, hjust = 1),
                   axis.text.y = element_text(size = 10))
    o <- o + labs(x = xlabel,
                  y = ylabel,
                  linetype = grouplabel)
    
    ## ADD A NOTE CONTAINING CRISIS PERIODS INCLUDED IN THE PLOT
    o <- arrangeGrob(o,
                     sub =
                         textGrob(paste0("Crisis episodes: ",episodes,"."),
                                  x = 0,
                                  hjust = 0,
                                  vjust=0.1,
                                  gp = gpar(fontface = "italic", fontsize = 5)))
    
    ## o <- o + theme(legend.position="none")
    return(o)
}

#https://github.com/hadley/ggplot2/wiki/Share-a-legend-between-two-ggplot2-graphs
.grabLegend <- function(plot){
    tmp <- ggplot_gtable(ggplot_build(plot))
    leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
    legend <- tmp$grobs[[leg]]
    return(legend)
}


##' .. content for \description{} (no empty lines) ..
##'
##' .. content for \details{} ..
##' @title Density plots of benchmark indicators for different crisis groups 
##' @param crisisType Type of crisis used in grouping observations
##' @param filename path to the resulting pdf file with graph output
##' @param adjust Adjust for time trends?
##' @return NULL
##' @author Janko Cizel
plotDensityAroundCrisisEvents <- function(crisisdb = loadCrisisDB(),
                                          crisisType = "debtcrisis",
                                          filename = '~/Downloads/test.pdf',
                                          adjust = FALSE,
                                          width = 320,
                                          height = 420,
                                          dtList =
                                              list("alt" = getAltmanZscore()),
                                          plotDefinition =
                                          list('ratingnum' =
                                               list(x = 'ratingnum',
                                                    xlabel = 'S&P Sovereign Credit Rating'),
                                               'cds' =
                                               list(x = 'cds',
                                                    xlabel = '5-Year Sovereign CDS Spread'),
                                               'spread' =
                                               list(x = 'spread',
                                                    xlabel = 'Sovereign Bond Yield Spread')),
                                          groups =
                                              list("[-4:-1]"=expression(COUNTDOWN %between% c(-4,-1)),
                                                   "[0]"=expression(COUNTDOWN == 0),
                                                   "[1:4]"=expression(COUNTDOWN %between% c(1,4)))){
    dt <- augmentBenchmarkDataset(crisisdb = crisisdb,
                                  dtList = dtList)

    dt1 <- 
        createCrisisVariables(crisisDT = dt,
                              crisisCol = crisisType,
                              idCol = "iso3",
                              timeCol = "year",
                              groups = groups)
    

    if (adjust == TRUE){
        cols <- sapply(plotDefinition, function(x) x$x)

        dt1[, paste(cols) := lapply(.SD, function(x){
            o <- (x-mean(x, na.rm = TRUE))/sd(x, na.rm = TRUE)
            return(o)
        }), by = year, .SDcols = cols]
    }

    setnames(dt1,names(dt1),.removeSpecialCharacters(names(dt1)))
    
    plot_list <- list()

    for (x in 1:length(plotDefinition)){
        plot_list[[x]] <- try({
            .t <- plotDefinition[[x]]
            o <- .densityPlot(data = dt1,
                              x = .removeSpecialCharacters(.t$x),
                              xlabel = .t$xlabel)
            ## o <- o + scale_x_continuous(limits = c(0,10000))
            o
        })
    }
    
    plot_list <- Filter(function(x) !"try-error" %in% class(x), plot_list)

    ## l <- .grabLegend(plot_list[[1]])

    g <- 
        do.call("arrangeGrob",
                c(plot_list,
                  list(
                      ## legend = legend,
                      as.table = FALSE,
                      ncol = 1
                      )
                  ))

    ggsave(filename = filename,
           width = width,
           height = height,
           units = "mm",
           plot = g)
    
    return(NULL)
}

## plotDensityAroundCrisisEvents(crisisType = "Sovereign Debt Crisis",
##                               filename = './inst/RESULTS/plotSovereignBenchmarkDens.pdf',
##                               adjust = TRUE)

## plotDensityAroundCrisisEvents(crisisType = "Sovereign Debt Crisis",
##                               adjust = TRUE)



.tsBoxPlot <- function(data,
                       x = 'as.factor(COUNTDOWN)',
                       y,
                       limits = c(-5,5),
                       xlabel = "",
                       ylabel = ""){
    episodes <- .createNoteOnCrisisEpisodes(data = data[!is.na(get(y))])
    
    d <- data[COUNTDOWN %between% limits]

    ## compute lower and upper whiskers
    ## .ylim = boxplot.stats(data[[y]])$stats[c(1, 5)]

    o <- ggplot(data = d)
    o <- o + geom_boxplot(aes_string(x = x,
                                     y = y))
    o <- o + theme_bw()
    o <- o + theme(axis.text.x = element_text(angle = 90, hjust = 1),
                   axis.text.y = element_text(size = 10))
    o <- o + labs(x = xlabel,
                  y = ylabel)
    ## o <- o + coord_cartesian(ylim = ylim1*1.05)

    ## ADD A NOTE CONTAINING CRISIS PERIODS INCLUDED IN THE PLOT
    o <- arrangeGrob(o,
                     sub =
                         textGrob(paste0("Crisis episodes: ",episodes,"."),
                                  x = 0,
                                  hjust = 0,
                                  vjust=0.1,
                                  gp = gpar(fontface = "italic", fontsize = 5)))    
    return(o)
}

## .tsBoxPlot(data = dt1,
##            y = 'ratingnum')


plotEventStudy <- function(
    crisisdb = loadCrisisDB(),
    crisisType = "debtcrisis",
    filename = '~/Downloads/test.pdf',
    adjust = FALSE,
    limits = c(-5,5),
    width = 320,
    height = 420,
    dtList =
        list("alt" = getAltmanZscore()),
    plotDefinition =
        list('ratingnum' =
                 list(y = 'ratingnum',
                      ylabel = 'S&P Sovereign Credit Rating'),
             'cds' =
                 list(y = 'cds',
                      ylabel = '5-Year Sovereign CDS Spread'),
             'spread' =
                 list(y = 'spread',
                      ylabel = 'Sovereign Bond Yield Spread')),
    groups =
        list("[-4:-1]"=expression(COUNTDOWN %between% c(-4,-1)),
             "[0]"=expression(COUNTDOWN == 0),
             "[1:4]"=expression(COUNTDOWN %between% c(1,4))))
{
    dt <- augmentBenchmarkDataset(crisisdb = crisisdb,
                                  dtList = dtList)

    dt1 <- 
        createCrisisVariables(crisisDT = dt,
                              crisisCol = crisisType,
                              idCol = "iso3",
                              timeCol = "year",
                              groups = groups)
    

    if (adjust == TRUE){
        cols <- sapply(plotDefinition, function(x) x$y)

        dt1[, paste(cols) := lapply(.SD, function(x){
            o <- (x-mean(x, na.rm = TRUE))/sd(x, na.rm = TRUE)
            return(o)
        }), by = year, .SDcols = cols]
    }

    setnames(dt1,names(dt1),.removeSpecialCharacters(names(dt1)))
    
    plot_list <- list()

    for (x in 1:length(plotDefinition)){
        plot_list[[x]] <- try({
            .t <- plotDefinition[[x]]
            o <- .tsBoxPlot(data = dt1,
                            y = .removeSpecialCharacters(.t$y),
                            ylabel = .t$ylabel,
                            limits = limits)
            ## o <- o + scale_x_continuous(limits = c(0,10000))
            o
        })
    }
    
    plot_list <- Filter(function(x) !"try-error" %in% class(x), plot_list)

    ## l <- .grabLegend(plot_list[[1]])

    g <- 
        do.call("arrangeGrob",
                c(plot_list,
                  list(
                      ## legend = legend,
                      as.table = FALSE,
                      ncol = 1
                  )
                  ))

    ggsave(filename = filename,
           width = width,
           height = height,
           units = "mm",
           plot = g)
    
    return(NULL)
}

## plotEventStudy(adjust = TRUE,
##                limits = c(-5,2))