CODEWITHSUNDEEP
rcolorsggplot2
user1738753
user1738753

Reputation: 636

Create different colours for regression lines

I am creating a scatterplot (around 1,500,000 points), and I running regressions through it based on a factor variable (see below "halfs"). This is the picture:

enter image description here

As you can see it is difficult to see the "red" regression lines.

This is the data:

  For_Cov     Per_chg      halfs
  1  0.8372001 0.002400000 upper half
  2  0.7236001 0.002800111 upper half
  3  0.6036000 0.000800000 upper half
  4  0.8540000 0.000000000 upper half
  5  0.9080001 0.003200000 upper half
  6  0.8248000 0.000000000 upper half
  7  0.1132000 0.000000000 upper half
  8  0.2044000 0.007600000 upper half
  9  0.2476001 0.085200000 upper half
  10 0.2368000 0.003600000 upper half

This is the code:

 ggplot(grid_gdp_full, aes(x = For_Cov, y = Def_Chg, group = factor(halfs))) + 
    geom_point(aes(colour = halfs), alpha = 0.1) +
    stat_smooth(method = "lm", formula = y ~ x + I(x^2), size = 4, fullrange=TRUE,     aes(group = halfs, colour = halfs), alpha = 1) +
    xlab("Initial") +
    ylab("Percent") +
    ylim(0,0.10) +
    scale_x_reverse() +
    theme_bw() +theme(
    axis.text=element_text(size=12)
    ,axis.title=element_text(size=14,face="bold")
    ,plot.background = element_blank()
    ,panel.grid.major = element_blank()
    ,panel.grid.minor = element_blank()
    ,panel.border = element_blank()
    ,panel.background = element_blank()
     ) +
     theme(axis.line = element_line(color = 'black'))

Does anyone how to change the code so that the regression lines will have different colors (regression line is created by stat_smooth) - possibly grey and blue for the two factor levels without changing the initial color of the points?

Here is the code I am using at the moment and I am still not able to get the two different colored lines:

 ggplot(grid_gdp_full, aes(x = For_Cov, y = Def_Chg, fill = factor(halfs))) + 
   geom_point(aes(colour = halfs), alpha = 0.1, colour="transparent",shape=21) +
   stat_smooth(method = "lm", formula = y ~ x + I(x^2), size = 2, fullrange=TRUE, aes(group = factor(halfs)), alpha = 1) +
   xlab("Initial") +
   ylab("Percent") +
   ylim(0,0.1) +
   scale_x_reverse() +
   theme_bw() +theme(
     axis.text=element_text(size=12)
     ,axis.title=element_text(size=14,face="bold")
     ,plot.background = element_blank()
     ,panel.grid.major = element_blank()
     ,panel.grid.minor = element_blank()
     ,panel.border = element_blank()
     ,panel.background = element_blank()
   ) +
   theme(axis.line = element_line(color = 'black')) +
   scale_colour_manual(values = c("red","blue")) + 
   scale_fill_manual(values = c("grey","green"))

dev.off()

Upvotes: 0

Views: 10812

Answers (2)

jlhoward
jlhoward

Reputation: 59355

OK. I see what you want now (I think...): you want the line groups to be a different color than the point groups. In essence you need four colors: two for point groups and two for the line groups.

This worked on your sample, recoded with For_Cov<0.5 to "lower" to split the data.

grid_gdp_full$halfs <- ifelse(grid_gdp_full$For_Cov<0.5,"lower half","upper half")
grid_gdp_full <- cbind(grid_gdp_full,col = 
                 ifelse(grid_gdp_full$halfs=="lower half","lm: lower","lm: upper"))

ggplot(grid_gdp_full, aes(x = For_Cov, y = Per_chg)) + 
  geom_point(aes(colour = factor(halfs)), alpha = 0.9, size=5) +
  stat_smooth(method = "lm", formula = y ~ x + I(x^2), size = 4, fullrange=TRUE,
              aes(group = col, colour = factor(col)), alpha = 1) +
  xlab("Initial") +
  ylab("Percent") +
  ylim(0,0.10) +
  scale_x_reverse() +
  theme_bw() +theme(
    axis.text=element_text(size=12)
    ,axis.title=element_text(size=14,face="bold")
    ,plot.background = element_blank()
    ,panel.grid.major = element_blank()
    ,panel.grid.minor = element_blank()
    ,panel.border = element_blank()
    ,panel.background = element_blank()
  ) +
  theme(axis.line = element_line(color = 'black'))

Produces this:

enter image description here

I made the points bigger and set alpha=0.9 so you can see the points in this small sample. Also you reversed the x axis, which makes "lower" and "upper" kind of confusing.

Upvotes: 2

user1317221_G
user1317221_G

Reputation: 15441

Taking advantage of point shape 21 we can use fill for points and colour for lines. Setting colours by scale_manual as we want. Note the colour=transparent in geom_point so we remove colour borders around points.

ggplot(mtcars, aes(x=cyl,y=mpg,fill=factor(gear))) +
            geom_point(aes(fill=factor(gear)),colour="transparent",shape=21) + 
            stat_smooth(aes(colour=factor(gear)),method="lm",se = FALSE) + 
            scale_colour_manual(values = c("red","blue", "green")) + 
            scale_fill_manual(values = c("orange","pink", "red"))

enter image description here

Upvotes: 4

Related Questions