How to hide a single key in R? - r

Question

Just to be clear: I am relatively new to R, and the code I am using is borrowed from someone else.
I have this graph for polling averages:
Here is my code: https://pastebin.com/qvQERRUH
library("tidyverse")
polls <- read.csv("polls_Paris.csv")
polls <- polls %>%
mutate(
date = format(as.Date(c(paste(year,month, day, sep="-")), by = "days"))
)
for(i in c("LFI", "PS", "EELV", "PP", "Griveaux", "LREM", "Villani", "Agir", "LR", "RN", "LP")) {
polls <- within(polls, {
assign(paste0("ci_", i), 1.96 * sqrt(( get(paste0("liste_", i)) * (100 - get(paste0("liste_", i)))) / n))
}
)
}
polls.10m <- polls[polls$date > seq(as.Date(Sys.Date()), length = 2, by = "-10 months")[2],]
polls.100 <- polls[order(as.Date(polls$date)),] %>% top_n(5000, as.Date(polls$date))
#Results = data.frame(date = as.Date("2019-12-01"), support = c(69.1,30.9))
svg('Opinion polling for the 2020 Paris municipal election.svg', width = 12, height = 6)
polls.100 %>%
gather(party, support, c(liste_LFI,liste_PS,liste_EELV,liste_PP,liste_Griveaux,liste_LREM,liste_Villani,liste_Agir,liste_LR,liste_RN,liste_LP), factor_key=TRUE) %>%
ggplot(aes(x=as.Date(date), y=support, colour=party)) +
geom_point(size=2.5, alpha=0.275) +
geom_smooth(se=FALSE, method="loess", span=1) +
labs(y = NULL,
x = NULL) +
guides(colour = guide_legend(ncol = 1, override.aes = list(linetype = 0, size = 3, alpha = 1))) +
scale_colour_manual(labels = c("Simonnet (LFI)", "Hidalgo (PS-PCF-G·s)", "Belliard (EELV)", "Gantzer (DVG)", "Griveaux (LREM-MR-UDI)", "Griveaux (avant diss. de Villani)", "Villani (Diss. LREM-PRG)", "Bournazel (Agir)", "Dati (LR)", "Federbusch (DVD-RN)", "Campion (SE)"), values = c("#cc2443", "#FF8080", "#00c000", "#ffc0c0", "#ffeb00", "#ffeb00", "#FF7F50", "#adc1fd", "#0066CC", "#0D378A", "#808080", "#808080")) +
theme(
plot.margin = margin(t = 0, unit = "cm"),
plot.background = element_blank(), panel.background = element_rect(fill = "grey92", colour = NA),
panel.border = element_blank(), legend.background = element_rect(fill = "transparent", colour = NA),
legend.key = element_rect(fill = "transparent", colour = NA), legend.title = element_blank(),
strip.background = element_rect(fill = "transparent", colour = NA),
panel.grid.major = element_line(colour = "#FFFFFF"), panel.grid.minor = element_line(colour = "#FFFFFF", size = 0.25),
axis.ticks = element_line(colour = "grey20"), axis.line = element_blank(),
plot.title = element_text(size = 12, hjust = 0),
plot.subtitle = element_text(size = 12, hjust = 0),
plot.caption = element_text(size = 12, colour = "#212121"),
axis.title = element_text(size = 12, face = "plain"), axis.text = element_text(size = 12, face = "plain", colour = "grey30"),
legend.position = "right",
legend.text = element_text(size = 12), strip.text = element_text(size = 12, face = "plain"),
legend.margin = margin(t = 0, unit = "cm"),
) +
scale_y_continuous(breaks = seq(0,33,5), minor_breaks = seq(0,33,1), limits = c(0, 33), expand = c(0, 0)) +
scale_x_date(breaks="6 months", minor_breaks="1 month", expand = c(0, 0))
#geom_point(data = Results, colour = c("#808080", "#E81B23"), size=4, shape=5) +
#geom_point(data = Results, colour = c("#808080", "#E81B23"), size=3.5, shape=18)
dev.off()
As you can see, Griveaux's line is split to separate the before-and-after of Villani's dissident candidacy; it's actually 2 separate lines (also separate in the dataset). Griveaux's name therefore has to appears twice.
How do I do to remove the key of a single set (remove the key for both the dots and regression line)?

Answer 1

Here is a hack. To remove a legend key, remove it from the breaks argument to scale_*_manual or equivalent but you must keep the same number of values as there are unique values in the color/fill aesthetic.
This is better shown with an example. I will use built-in data set iris.
To remove the legend key relative to "versicolor",
levels(df1$Species)
#[1] "setosa" "versicolor" "virginica"
just don't include it in the breaks.
library(ggplot2)
df1 <- iris[3:5]
ggplot(df1, aes(Petal.Length, Petal.Width, color = Species)) +
geom_point() +
geom_smooth(se = FALSE, method = "loess", span = 1) +
scale_color_manual(breaks = c("setosa", "virginica"),
values = c("red", "green", "blue"))