如何使用ggplot以不同颜色指示负值？

Question

I would like to display plot of a data frame which includes negative values either. In this case, the negative values may be indicated with red line. However I couldn't succeed. When try to draw the plot, the negatives values began to follow sequence because of cbind argument. In addition, I don't know another way to combine the data frames.

set.seed(123)
tmp <- data.frame(time = 1:10, quantity = round(runif(10, -15, 25)) )

neg <- tmp$quantity[tmp$quantity <0]
tmp.pred <-  cbind(tmp,neg)

y <- ggplot(tmp, aes(time, quantity)) + 
 geom_line() +   
 geom_line(data=tmp.pred, aes(y=neg),color="red") 
 y

Answer 1

A couple of methods I think get you close to what you're after.

library(ggplot2)

set.seed(123)
tmp <- data.frame(time = 1:10, quantity = round(runif(10, -15, 25)) )

method 1 - colour line by expression

This will colour the line from the point the quantity is negative, until the time the quantity is non-negative again.

ggplot(tmp, aes(time, quantity, colour=(quantity <= 0))) + 
geom_line(aes(group=1))

Method 2 - adding an extra column to so we can colour two lines, one is the complete 'quantity' column, and the other is only those values that are negative, or either size of a negative. (otherwise they'll be NA)

library(dplyr)
tmp <- tmp %>%
    mutate(neg = ifelse(quantity <= 0, quantity, NA)) %>%
    mutate(neg = ifelse(is.na(neg) & lag(neg) <= 0, quantity, neg)) %>%
    mutate(neg = ifelse(is.na(neg) & lead(neg) <= 0, quantity, neg))

ggplot() +
    geom_line(data=tmp, aes(x=time, y=quantity) , colour="black") +
    geom_line(data=tmp, aes(x=time, y=neg), colour="red")

If you want it to only change colour when the line crosses the x==0 line you may have to manipulate the data a bit more

Answer 2

Does this do what you want?

library(ggplot2)
set.seed(1)
tmp <- data.frame(time = 1:10, quantity = round(runif(10, -15, 25)) )
tmp$series <- ifelse(tmp$quantity < 0, "Negative", "Positive")
y <- ggplot(tmp, aes(x = time, y = quantity, colour = series)) + 
  geom_line() +
  scale_colour_manual(values=c("red", "blue"))
y