Adding a legend to a double plot using ggplot - r

Question

I'm trying to add a legend to my plot using ggplot in R. Everything OK so far. My case is special because I'm trying to deal with three variables, but not in order to draw a 3D plot but draw a 2D plot facing v1 vs. v2 and v1 vs. v3.
I get my plot in a correct way but I dont get the legend.
This is my code:
colfuncWarmest <- colorRampPalette(c("orange","red"))
colfuncColdest <- colorRampPalette(c("green","blue"))
plot <- ggplot(data=temperatures_Celsius, aes(x=temperatures_Celsius$Year))
params <- labs(title=paste("Year vs. (Warmest minimum temperature\n",
"and Coldest minimum temperature)"),
x="Year",
y="Coldest min temp / Warmest min temp")
theme <- theme(plot.title = element_text(hjust = 0.5)) #Centering title
wmtl<-geom_line(data=temperatures_Celsius,
aes(y=temperatures_Celsius$Warmest.Minimum.Temperature..C.,
color="red"
),
colour=colfuncWarmest(length(temperatures_Celsius$Year))
)
wmtt<-stat_smooth(data=temperatures_Celsius,
aes(y=temperatures_Celsius$Warmest.Minimum.Temperature..C.),
color="green",
method = "loess")
cmtl<- geom_line(data=temperatures_Celsius,
aes(y=temperatures_Celsius$Coldest.Minimum.Temperature..C.,
color="blue"
),
colour=colfuncColdest(length(temperatures_Celsius$Year))
)
cmtt<-stat_smooth(data=temperatures_Celsius,
aes(y=temperatures_Celsius$Coldest.Minimum.Temperature..C.),
color="orange",
method = "loess")
plot + theme + params + wmtl + wmtt + cmtl + cmtt
(Not all code was added because I did a lot of changes. It is only to get an idea) I get this:
If I add
+ scale_color_manual(values=c("red","blue"))
(for example) in order to add the legend, I get no error, but nothing different happens. I get the same plot.
What I want is only two lines. A red one that says "Warmest minimum" and another blue line that says "Coldest minimum". What could I do to get my legend in this way?
Thanks in advance.

Answer 1

Generally I would say that the correct way to apply a legend to a ggplot is to map a variable to an aesthetic (such as fill, color, size, alpha). Usually this consists of transforming the data to long format (key ~ value pair) and mapping the key variable to color or other aestetic.
In the current case this is not desirable since there is next to no chance the color gradient (colorRampPalette) on the line could be achieved. So I suggest a hacky way where a dummy layer (layer which will not be seen on the plot) is used to create the legend.
Here is some data
temperatures_Celsius = data.frame(year = 1900:2000,
Warmest = rnorm(100, mean = 20, sd = 5),
Coldest = rnorm(100, mean = 10, sd = 5))
Your plot:
colfuncWarmest <- colorRampPalette(c("orange","red"))
colfuncColdest <- colorRampPalette(c("green","blue"))
plot <- ggplot(data=temperatures_Celsius, aes(x=year))
params <- labs(title=paste("Year vs. (Warmest minimum temperature\n",
"and Coldest minimum temperature)"),
x="Year",
y="Coldest min temp / Warmest min temp")
theme <- theme(plot.title = element_text(hjust = 0.5)) #Centering title
wmtl<-geom_line(data=temperatures_Celsius,
aes(y=Warmest),
colour=colfuncWarmest(length(temperatures_Celsius$year)))
wmtt<-stat_smooth(data=temperatures_Celsius,
aes(y=Warmest),
color="green",
method = "loess")
cmtl<- geom_line(data=temperatures_Celsius,
aes(y=Coldest),
colour=colfuncColdest(length(temperatures_Celsius$year)))
cmtt<-stat_smooth(data=temperatures_Celsius,
aes(y=Coldest),
color="orange",
method = "loess")
plot1 <- plot + theme + params + wmtl + wmtt + cmtl + cmtt
Now add a dummy layer:
plot1+
geom_line(data = data.frame(year = c(1900, 1900),
group = factor(c("Coldest", "Warmest"), levels = c("Warmest", "Coldest")),
value = c(10, 20)), aes(x=year, y = value, color = group), size = 2)+
scale_color_manual(values=c("red","blue"))