Plot time variations — gg_timevariation • rOstluft.plot

Inspired by openair::timeVariation() multiple functions to plot diurnal, day hour of the week, day of the week and monthly variation. The functions are optimized for data in long format and the defaults are for the rolf format from the rOstluft Package.

Usage

gg_timevariation(
  data,
  dt = "starttime",
  y = "value",
  group = NULL,
  id_cols = grp("site", "parameter", "interval", "unit"),
  statistic = c("mean", "median"),
  draw_ci = TRUE,
  conf_interval = NULL,
  B = 1000,
  ylab = ggplot2::waiver(),
  ylim = c(NA, NA),
  legend_title = NULL,
  language_code = "de",
  geom = ggplot2::geom_line(size = 1),
  geom_ci = ggplot2::geom_ribbon(alpha = 0.2)
)

gg_timevariation_wday_hour(
  data,
  dt = "starttime",
  y = "value",
  group = NULL,
  id_cols = grp("site", "parameter", "interval", "unit"),
  statistic = c("mean", "median"),
  draw_ci = TRUE,
  conf_interval = NULL,
  ylab = ggplot2::waiver(),
  B = 1000,
  ylim = c(NA, NA),
  legend_title = NULL,
  language_code = "de",
  geom = ggplot2::geom_line(size = 1),
  geom_ci = ggplot2::geom_ribbon(alpha = 0.2)
)

gg_timevariation_wday(
  data,
  dt = "starttime",
  y = "value",
  group = NULL,
  id_cols = grp("site", "parameter", "interval", "unit"),
  statistic = c("mean", "median"),
  draw_ci = TRUE,
  conf_interval = NULL,
  B = 1000,
  ylab = ggplot2::waiver(),
  ylim = c(NA, NA),
  legend_title = NULL,
  language_code = "de",
  geom = ggplot2::geom_line(size = 1),
  geom_ci = ggplot2::geom_ribbon(alpha = 0.2)
)

gg_timevariation_month(
  data,
  dt = "starttime",
  y = "value",
  group = NULL,
  id_cols = grp("site", "parameter", "interval", "unit"),
  statistic = c("mean", "median"),
  draw_ci = TRUE,
  conf_interval = NULL,
  B = 1000,
  ylab = ggplot2::waiver(),
  ylim = c(NA, NA),
  legend_title = NULL,
  language_code = "de",
  geom = ggplot2::geom_line(size = 1),
  geom_ci = ggplot2::geom_ribbon(alpha = 0.2)
)

gg_timevariation_diurnal(
  data,
  dt = "starttime",
  y = "value",
  group = NULL,
  id_cols = grp("site", "parameter", "interval", "unit"),
  statistic = c("mean", "median"),
  draw_ci = TRUE,
  conf_interval = NULL,
  B = 1000,
  ylab = ggplot2::waiver(),
  ylim = c(NA, NA),
  legend_title = NULL,
  language_code = "de",
  geom = ggplot2::geom_line(size = 1),
  geom_ci = ggplot2::geom_ribbon(alpha = 0.2)
)

Arguments

data: data frame with input data as hourly time series
dt: a string or symbol for the date time column (default: starttime)
y: a string or symbol specifying the target column to be summarised (default: value)
group: column as string to be used to split the input data in multiple groups. Should be a member of id_cols eg. parameter or site (default NULL)
id_cols: A set of columns that uniquely identifies each observation. Use rOstluft.plot::grp() for quoting. default rOstluft.plot::grp(site, parameter, interval, unit)
statistic: Can be “mean” (default) or “median”. If the statistic is ‘mean’ then the mean line and the 95% confidence interval in the mean are plotted by default. ggplot2::mean_cl_boot() is used to calculated the intervals trough bootstrap simulations without assuming normality. If the statistic is ‘median’ then the median line is plotted together with the 25/75th quantiles are plotted. Users can control the confidence intervals with draw_ci and conf_interval
draw_ci: if TRUE draw confidence interval using ggplot2::mean_cl_boot() for "mean" and ggplot2::median_hilow() for "median".
conf_interval: for "mean" (ggplot2::mean_cl_boot()) specifies the confidence level (0-1) for interval estimation of the population mean. For "median" (ggplot2::median_hilow()) conf_interval is the coverage probability the outer quantiles should target. When the default, 0.5, is used, the lower and upper quantiles computed are 0.25 and 0.75
B: number of bootstrap resamples for ggplot2::mean_cl_boot()
ylab: provide a custom y plot label
ylim: limits for y scale see ggplot2::scale_y_continuous() for more infos.
legend_title: provide a legend title
language_code: ISO country code for the language used as weekdays and months labels (default: "de")
geom: geom used for rendering default ggplot2::geom_line()
geom_ci: geom used for rendering confidence interval. Must support ymin/ymax mapping. default ggplot2::geom_ribbon()

Value

a ggplot2::ggplot() object or in case of gg_timevariaton() a patchwork::patchwork object

Examples

library(ggplot2)

fn <- rOstluft.data::f("Zch_Stampfenbachstrasse_2010-2014.csv")

data <-
  rOstluft::read_airmo_csv(fn) %>%
  rOstluft::pluck_parameter("NOx", "NO", "NO2") %>%
  rOstluft::resample(new_interval = "h1")

# monthly variation of data
gg_timevariation_month(data, group = "parameter", ylab = "Konzentration [µg/m3]")
#> plotting with statistic mean and confidence interval of 0.95


# don't draw a confidence interval
gg_timevariation_month(data, group = "parameter", draw_ci = FALSE, ylab = "Konzentration [µg/m3]")


# for faceting the variable must be included in the id_cols
gg_timevariation_wday(
  data,
  group = "parameter",
  id_cols = grp(site, parameter, interval, unit, season = cut_season(starttime))
) + facet_wrap(vars(season))
#> plotting with statistic mean and confidence interval of 0.95


# utility function to compose all plots together using patchwork
# for advanced use cases you should probably compose the plot yourself
# you can use ylim to start all y axis by 0. Lowering B speed up the process.
gg_timevariation(
  data,
  group = "parameter",
  B = 10,
  ylab = "Konzentration [µg/m3]",
  ylim = c(0, NA)
)
#> plotting with statistic mean and confidence interval of 0.95
#> plotting with statistic mean and confidence interval of 0.95
#> plotting with statistic mean and confidence interval of 0.95
#> plotting with statistic mean and confidence interval of 0.95



# you can use wide data, but only with a single pollutant
data_wide <- rOstluft::rolf_to_openair(data)

gg_timevariation_wday_hour(
  data_wide,
  dt = date,
  y = NO2,
  group = "site",
  id_cols = grp(site)
)
#> plotting with statistic mean and confidence interval of 0.95


# you can also use a function in id_cols to create groups
gg_timevariation_diurnal(
  data_wide,
  dt = date,
  y = NO2,
  group = "season",
  id_cols = grp(site, season = cut_season(date))
)
#> plotting with statistic mean and confidence interval of 0.95


# use an alternative geom
gg_timevariation_month(
  data,
  group = "parameter",
  ylab = "Konzentration [µg/m3]",
  B = 10,
  geom_ci = geom_linerange(size = 2)
) +
scale_x_discrete(
  expand = expansion(mult = 0.02)
)
#> plotting with statistic mean and confidence interval of 0.95
#> Scale for x is already present.
#> Adding another scale for x, which will replace the existing scale.