The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.
library(nomisdata)
library(ggplot2)
library(dplyr)
# Fetch unemployment data over time
unemployment <- fetch_nomis(
"NM_1_1",
date = paste0("2020", c("01", "02", "03", "04", "05", "06")),
geography = "2092957697", # UK
measures = 20201, # Rate
sex = 7
)
# Plot
unemployment |>
mutate(DATE = as.Date(paste0(DATE, "-01"))) |>
ggplot(aes(x = DATE, y = OBS_VALUE)) +
geom_line(linewidth = 1, colour = "#0066cc") +
geom_point(size = 2, colour = "#0066cc") +
labs(
title = "UK Unemployment Rate",
subtitle = "Claimant Count Rate, 2020",
x = NULL,
y = "Rate (%)",
caption = "Source: Nomis / ONS"
) +
theme_minimal() +
theme(
plot.title = element_text(face = "bold", size = 14),
panel.grid.minor = element_blank()
)# Fetch data for all regions
regions <- fetch_nomis(
"NM_1_1",
time = "latest",
geography = "TYPE480", # Regions
measures = 20201, # Rate
sex = 7
)
# Bar chart
regions |>
arrange(desc(OBS_VALUE)) |>
ggplot(aes(x = reorder(GEOGRAPHY_NAME, OBS_VALUE), y = OBS_VALUE)) +
geom_col(fill = "#0066cc", alpha = 0.8) +
coord_flip() +
labs(
title = "Unemployment Rate by Region",
x = NULL,
y = "Claimant Count Rate (%)",
caption = "Source: Nomis / ONS"
) +
theme_minimal()# Fetch data by sex
by_sex <- fetch_nomis(
"NM_1_1",
time = "latest",
geography = "TYPE499", # Countries
measures = 20201,
sex = c(5, 6) # Male, Female
)
# Grouped bar chart
by_sex |>
ggplot(aes(x = GEOGRAPHY_NAME, y = OBS_VALUE, fill = SEX_NAME)) +
geom_col(position = "dodge", alpha = 0.8) +
scale_fill_manual(values = c("#0066cc", "#cc0066")) +
labs(
title = "Unemployment Rate by Sex and Country",
x = NULL,
y = "Rate (%)",
fill = "Sex",
caption = "Source: Nomis / ONS"
) +
theme_minimal() +
theme(legend.position = "top")library(sf)
# Fetch spatial data
spatial_data <- fetch_spatial(
"NM_1_1",
time = "latest",
geography = "TYPE480",
measures = 20201,
sex = 7
)
# Map
ggplot(spatial_data) +
geom_sf(aes(fill = OBS_VALUE), colour = "white", size = 0.3) +
scale_fill_viridis_c(option = "plasma") +
labs(
title = "Unemployment Rate by Region",
fill = "Rate (%)"
) +
theme_void()These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.