Plots for many series over time

Base-graphics plots in the house style for a value observed by period, for one series or many:

plotcal()

a calendar heatmap of a daily series: one tile per day, weeks across and weekdays down, one block per year.

plothrz()

a horizon chart: each series in its own thin row, its deviation from its own median folded into bands shaded bands, above in blue and below in red, so many series fit in little height.

plotsm()

small multiples: one panel per series on a shared scale.

plotbump()

a bump chart: the rank of each series in each period (1 = largest), so changes of order stand out.

Usage

plotcal(x, y, ..., main = NULL, col = NULL, pal = NULL)

plothrz(
  y,
  prd,
  grp = NULL,
  bands = 3L,
  ...,
  main = NULL,
  col = NULL,
  pal = NULL
)

plotsm(y, prd, grp = NULL, ..., main = NULL, col = NULL, pal = NULL)

plotbump(y, prd, grp, desc = TRUE, ..., main = NULL, col = NULL, pal = NULL)

Arguments

x

Dates (a Date or text as.Date() reads) for plotcal().

y

Numeric vector of values.

...

Further arguments to the plot() call that opens the frame.

main, col, pal

The title, colours overriding the palette, and a blpal() name.

prd

Vector of periods, one per value (numbers, dates or text in order of appearance).

grp

Vector of series names, one per value, or NULL for one series.

bands

Whole number, default 3: the bands of a horizon chart.

desc

Logical, default TRUE: rank 1 is the largest value (FALSE: the smallest).

Value

The data drawn, invisibly: for plotcal() the days (date, year, week, wday, y), for the others the series (grp, prd, y, and rank for plotbump(), band for plothrz()).

Examples

set.seed(5)
days <- seq(as.Date("2024-01-01"), as.Date("2024-12-31"), by = "day")
plotcal(days, stats::rpois(length(days), 5 + 3 * (format(days, "%u") > "5")))

d <- expand.grid(
  year = 2015:2024,
  area = c("north", "south", "east", "west")
)
d$n <- stats::rpois(
  nrow(d),
  50 + 10 * as.integer(d$area) + 2 * (d$year - 2015)
)
plotsm(d$n, d$year, d$area)

plothrz(d$n, d$year, d$area)

plotbump(d$n, d$year, d$area)

try(plotcal("not a date", 1))
#> Error : plotcal: `x` must be dates, one per value of the numeric `y`