Ate Poorthuis 02/09/2019
To start creating our figures, we read in two datasets: a spatial datafile (in rds/sf format, but can read in from geojson or similar as well) and an attribute table containing all the necessary counts and metrics per hexagon. We join the two together based on their common id attribute.
hex_data <- readRDS(here("analysis/data/derived_data/global_hex_join_odds_ratios.rds"))
hex_sf <- readRDS(here("analysis/data/derived_data/global_hex_grid.rds")) %>%
left_join(., hex_data, by = "hex")
hex_union <- hex_sf %>%
st_buffer(1) %>%
summarise(count = n())We will create the figures with ggplot. To make them look a little cleaner than the default theme, we overwrite some theme settings. These are mostly global settings but we'll overwrite them as-needed for some specific figures later.
basecolor <- "#FFF5EB"
colors <- c("#FFF5EB", "#FDD1A5", "#FD9243", "#DE4F05", "#7F2704")
breaks <- c(0, 100, 1000, 10000, 50000, 1000000)
labels <- c(" 0", "100", "1,000", "10,000", "50,000+")
font_col <- "#22211d"
## theme
map_theme <- theme(text = element_text(family = "Helvetica Neue", color = font_col),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.border = element_blank(),
panel.background = element_blank(),
axis.text.x=element_blank(),
axis.title.x=element_blank(),
axis.ticks.x=element_blank(),
axis.text.y=element_blank(),
axis.title.y=element_blank(),
axis.ticks.y=element_blank(),
plot.title=element_text(hjust = 0, family = "Helvetica Neue", color = font_col, angle = 0, size = 12),
plot.subtitle=element_text(family = "Helvetica Neue Thin Italic", color = font_col, angle = 0, size = 10),
plot.caption=element_text(family = "Helvetica Neue Thin", color = font_col, angle = 0, size = 7),
legend.background = element_rect(fill="transparent"),
legend.position = c(0.675, 0.08),
legend.box = "horizontal")
## legend
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = font_col, angle = 0, size = 12),
title = "Count (absolute intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = font_col, angle = 0, size = 10),
nrow = 1,
byrow = T)
## color scale
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)The total attention to fashion world-wide - based on absolute count for all keywords.
fig2 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(total, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Fashion", subtitle = "absolute counts for all keywords")
fig2 <- geographyoffashion::annotate_legend(fig2, label1 = "more attention", label2 = "less attention")
plot(fig2)
#ggsave(plot = fig2, filename = "Figure2.pdf", width = 9, height = 5.5, dpi = 600, device = cairo_pdf)The relative intensity/attention to fashion world-wide - based on the odds ratio. N.B. we use a random global sample of tweets to calculate the odds ratio here (see manuscript for details).
breaks <- c(0, 0.5, 0.8, 1/0.8, 1/0.5, 100000000)
labels <- c("0", "0.5", "0.8", "1.25", "2+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Odds Ratio (relative intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
fig4 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(total_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Fashion", subtitle = "odds ratio for all keywords")
fig4 <- geographyoffashion::annotate_legend(fig4, label1 = "more attention", label2 = "less attention")
plot(fig4)
#ggsave(plot = fig4, filename = "Figure4.png", width = 9, height = 5.5, dpi = 600)Total attention to “creatives,” “business,” and “marketing” subsectors of the fashion industry (odds ratio).
p1 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(creatives_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Creatives")
p2 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(business_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Business")
p3 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(marketing_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Marketing")
g <- geographyoffashion::grid_arrange_shared_legend(p1, p2, p3)
plot(g)
#ggsave(plot = g, filename = "Figure5.png", width = 9, height = 16.5, dpi = 300)Shannon diversity index of fashion attention.
breaks <- c(0, 1, 2, 3, 3.5, 100000000)
labels <- c("0", "1", "2", "3", "3.5")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Diversity (Shannon index)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
fig6 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(divShannon, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Fashion", subtitle = "diversity for all keywords")
fig6 <- geographyoffashion::annotate_legend(fig6, label1 = "more diversity", label2 = "less diversity")
plot(fig6)
#ggsave(plot = fig4, filename = "Figure6.png", width = 9, height = 5.5, dpi = 600)Total attention by number of followers. (A) Absolute counts for all keywords and (B) odds ratios for all keywords.
breaks <- c(0, 0.5, 0.8, 1/0.8, 1/0.5, 100000000)
labels <- c("0", "0.5", "0.8", "1.25", "2+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Odds Ratio (relative intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p1 <- hex_sf %>%
filter(total > 10) %>%
filter(total_fol > 0) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(total_fol_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention by Number of Followers", subtitle = "odds ratio for all keywords")
p1 <- geographyoffashion::annotate_legend(p1, label1 = "more attention", label2 = "less attention")
breaks <- c(0, 10000, 100000, 1000000, 10000000, 21637898124)
labels <- c("0", "10k", "100k", "1M", "10M+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Count (absolute intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p2 <- hex_sf %>%
filter(total > 10) %>%
filter(total_fol > 0) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(total_fol, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention by Number of Followers", subtitle = "absolute count for all keywords")
p2 <- geographyoffashion::annotate_legend(p2, label1 = "more attention", label2 = "less attention")
fig7 <- cbind(p2, p1, size = "first")
plot(fig7)
#ggsave(plot = fig7, filename = "Figure7.png", width = 18, height = 5.5, dpi = 600)Total attention to Italian fashion. (A) Absolute counts for all keywords and (B) odds ratios for all keywords.
breaks <- c(0, 0.5, 0.8, 1/0.8, 1/0.5, 100000000)
labels <- c("0", "0.5", "0.8", "1.25", "2+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Odds Ratio (relative intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p1 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(NAT_ITALY_ORc, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Italian Fashion")
p1 <- geographyoffashion::annotate_legend(p1, label1 = "more attention", label2 = "less attention")
breaks <- c(0, 25, 50, 100, 500, 10000000000000)
labels <- c("0", "25", "50", "100", "500+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Count (absolute intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p2 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(NAT_ITALY, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Italian Fashion")
p2 <- geographyoffashion::annotate_legend(p2, label1 = "more attention", label2 = "less attention")
fig8 <- cbind(p2, p1, size = "first")
plot(fig8)
#ggsave(plot = fig8, filename = "Figure8.png", width = 18, height = 5.5, dpi = 600)Total attention to Gucci. (A) Absolute counts and (B) odds ratio.
breaks <- c(0, 0.5, 0.8, 1/0.8, 1/0.5, 100000000)
labels <- c("0", "0.5", "0.8", "1.25", "2+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Odds Ratio (relative intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p1 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(`Associated_companies-040_ORc`, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Gucci")
p1 <- geographyoffashion::annotate_legend(p1, label1 = "more attention", label2 = "less attention")
breaks <- c(0, 25, 50, 100, 500, 10000000000000)
labels <- c("0", "25", "50", "100", "500+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Count (absolute intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p2 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(`Associated_companies-040`, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Gucci")
p2 <- geographyoffashion::annotate_legend(p2, label1 = "more attention", label2 = "less attention")
fig9 <- cbind(p2, p1, size = "first")
plot(fig9)
#ggsave(plot = fig9, filename = "Figure9.png", width = 18, height = 5.5, dpi = 600)Total attention to Louis Vuitton (A) Absolute counts and (B) odds ratio.
breaks <- c(0, 0.5, 0.8, 1/0.8, 1/0.5, 100000000)
labels <- c("0", "0.5", "0.8", "1.25", "2+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Odds Ratio (relative intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p1 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(`Associated_companies-056_ORc`, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Louis Vuitton")
p1 <- geographyoffashion::annotate_legend(p1, label1 = "more attention", label2 = "less attention")
breaks <- c(0, 25, 50, 100, 500, 10000000000000)
labels <- c("0", "25", "50", "100", "500+")
legend <- guide_legend(keyheight = unit(4, units = "mm"),
keywidth = unit(70 / length(labels), units = "mm"),
direction = "horizontal", title.position = "top",
label.position="top",
title.theme = element_text(family = "Helvetica Neue Medium", color = "#22211d", angle = 0, size = 12),
title = "Count (absolute intensity)",
title.hjust = 0.0,
label.hjust = -0.2,
label.theme = element_text(family = "Helvetica Neue Thin", color = "#22211d", angle = 0, size = 10),
nrow = 1,
byrow = T)
scale_fill <- scale_fill_manual(values=colors, guide = legend, labels = labels)
p2 <- hex_sf %>%
filter(total > 10) %>%
ggplot() +
ylim(c(-6000148, 8248865)) + xlim(c(-10097010, 14597010)) +
geom_sf(data = hex_union, fill = basecolor, color = NA) +
geom_sf(aes(fill = cut(`Associated_companies-056`, breaks=breaks, include.lowest = T)), color = NA) +
scale_fill +
geom_sf(data = hex_union, fill = NA, color = "gray75", size = 0.1) +
coord_sf(datum = NA) +
map_theme +
labs(title = "Total Attention to Louis Vuitton")
p2 <- geographyoffashion::annotate_legend(p2, label1 = "more attention", label2 = "less attention")
fig10 <- cbind(p2, p1, size = "first")
plot(fig10)
#ggsave(plot = fig10, filename = "Figure10.png", width = 18, height = 5.5, dpi = 600)