How do I assign circle and period shapes in ggplot2? Right now I can do two shapes or two symbols but not both.
data.frame(
x = rnorm(10),
y = rnorm(10),
group = gl(2, 5, labels = c("circle", "period"))
) %>%
ggplot(aes(x = x, y = y, shape = group)) +
geom_point(size = 4) +
# scale_shape_manual(values = c("a", ".")) + # okay
# scale_shape_manual(values = c("circle", "square")) + # okay
scale_shape_manual(values = c("circle", ".")) # error
Based on this comment I don't think you can mix symbols and number references. But (the same comment) shows all possible shapes and their corresponding numbers. The period is 46 and circle is 1.
Using the numbers for the shapes you want:
data.frame(
x = rnorm(10),
y = rnorm(10),
group = gl(2, 5, labels = c("circle", "period"))
) %>%
ggplot(aes(x = x, y = y, shape = group)) +
geom_point(size = 4) +
scale_shape_manual(values = c(1, 46))
I like using a named character vector to define the manual shapes for each group.
plotting_shapes <- c("group_1" = 1,
"group_2" = 46)
data.frame(
x = rnorm(10),
y = rnorm(10),
group = gl(2, 5, labels = c("group_1", "group_2"))
) %>%
ggplot(aes(x = x, y = y, shape = group)) +
geom_point(size = 4) +
scale_shape_manual(values = plotting_shapes)
And so if you really wanted to use "." to refer to shape 46, then you could achieve this through the following:
plotting_shapes <- c("circle" = 1,
"." = 46)
data.frame(
x = rnorm(10),
y = rnorm(10),
group = gl(2, 5, labels = c("group_1", "group_2"))
) %>%
ggplot(aes(x = x, y = y, shape = group)) +
geom_point(size = 4) +
scale_shape_manual(values = plotting_shapes[c("circle", ".")] %>%
unname()) # unname() is required because
# the names don't correspond to
# the group names "group_1" and "group_2"
Related
The problem is similar to this question but here the two layers use different geometries, geom_tile and geom_point. The idea is to have tiles show up at different locations only in frames 2, 5, 8, and the point move along the diagonal in every frame.
When trying to run the following example, I get the error:
Error: time data must be the same class in all layers
Example
require(data.table)
require(ggplot2)
require(gganimate)
# 3 tiles along x = 10-y; present at time points 2, 5, 8
dtP1 = data.table(x = c(1, 5, 9),
y = c(9, 5, 1),
t = c(2, 5, 8))
# 9 points along x=y; present at every time point
dtP2 = data.table(x = 1:9,
y = 1:9,
t = 1:9)
p = ggplot() +
geom_tile(data = dtP1,
aes(x = x,
y = y),
color = "#000000") +
geom_point(data = dtP2,
aes(x = x,
y = y),
color = "#FF0000") +
gganimate::transition_time(t) +
gganimate::ease_aes('linear')
pAnim = gganimate::animate(p,
renderer = av_renderer("~/test.mp4"),
fps = 1,
nframes = 9,
height = 400, width = 400)
Does the following work for you?
library(dplyr)
p <- rbind(dtP1 %>% mutate(group = "group1"),
dtP2 %>% mutate(group = "group2")) %>%
tidyr::complete(t, group) %>%
ggplot(aes(x = x, y = y)) +
geom_tile(data = . %>% filter(group == "group1"),
color = "black") +
geom_point(data = . %>% filter(group == "group2"),
color = "red") +
ggtitle("{frame_time}") + # added this to show the frame explicitly; optional
transition_time(t) +
ease_aes('linear')
animate(p, nframes = 9, fps = 1)
I'm plotting a discrete CDF. I have a few questions regarding geom_step which I'm not finding by using Google.
Is it possible to make the line segment representing the jump dashed
rather than solid to better show whats going on?
Is it possible to add geom_point more efficiently than I do? (less
c/p).
Below is my current solution:
library(tidyverse)
library(ggthemes)
theme_set(theme_few())
x0 <- seq(-0.5, -0.01, by = 0.01)
x1 <- seq(0, 0.99, by = 0.02)
x2 <- seq(1, 1.99, by = 0.02)
x3 <- seq(2, 2.99, by = 0.02)
x35 <- seq(3, 3.49, by = 0.01)
x4 <- seq(3.5, 3.99, by = 0.01)
tibble_ex <- tibble(
x0 = x0,
x1 = x1,
x2 = x2,
x3 = x3,
x35 = x35,
x4 = x4
)
tibble_ex %>%
gather(x, xax, x0:x4) %>%
mutate(cdf = case_when(x == 'x0' ~ 0,
x == 'x1' ~ 1/2,
x == 'x2' ~ 3/5,
x == 'x3' ~ 4/5,
x == 'x35' ~ 9/10,
x == 'x4' ~ 1)) %>%
ggplot(aes(x = xax, y = cdf)) +
geom_step() +
geom_point(aes(x = 0, y = 0), size = 3, shape = 21, fill = 'white') +
geom_point(aes(x = 1, y = 0.5), size = 3, shape = 21, fill = 'white') +
geom_point(aes(x = 2, y = 3/5), size = 3, shape = 21, fill = 'white') +
geom_point(aes(x = 3, y = 4/5), size = 3, shape = 21, fill = 'white') +
geom_point(aes(x = 3.5, y = 9/10), size = 3, shape = 21, fill = 'white') +
geom_point(aes(x = 0, y = 0.5), size = 3, shape = 21, fill = 'black') +
geom_point(aes(x = 1, y = 3/5), size = 3, shape = 21, fill = 'black') +
geom_point(aes(x = 2, y = 4/5), size = 3, shape = 21, fill = 'black') +
geom_point(aes(x = 3, y = 9/10), size = 3, shape = 21, fill = 'black') +
geom_point(aes(x = 3.5, y = 1), size = 3, shape = 21, fill = 'black') +
labs(x = 'x', y = 'F(x)')
ggplot will be more powerful to use if you can put your data into a data frame and structure it so that the characteristics of your data can be mapped directly.
Here's a way to take your data and augment it with additional rows that represent the connecting points, by matching each x with the prior cdf value. I added a column, type, to keep track of which is which. I also arrange df so that geom_segment plots the points in the right order.
new_steps <-
tibble(x = c(0:3, 3.5, 4),
cdf = c(0, .5, .6, .8, .9, 1))
df <- new_steps %>%
mutate(type = "cdf") %>%
bind_rows(new_steps %>%
mutate(type = "prior",
cdf = lag(cdf))) %>%
drop_na() %>%
arrange(x, desc(type))
Then we can map the points' fill and the geom_segments' linetype to type.
ggplot(df) +
geom_point(aes(x, cdf, fill = type),
shape = 21) +
scale_fill_manual(values = c("black", "white")) +
geom_segment(aes(x = lag(x), y = lag(cdf),
xend = x, yend = cdf,
lty = type)) +
scale_linetype_manual(values = c("dashed", "solid"))
(1) No, there is not a built-in way to make the geom_step half-dashed. But if you post this as a separate question, perhaps someone will help create a new geom for this.
(2) The answer is to put the points you want plotted in a data frame, like anything else you might want to plot:
point_data = data.frame(x = rep(c(0, 1, 2, 3, 3.5), 2),
y = c(0, rep(c(.5, .6, .8, .9), 2), 1),
z = rep(c("a", "b"), each = 5))
# calling your gathered/mutated version of tibble_ex df
ggplot(df, aes(x = xax, y = cdf)) +
geom_step() +
geom_point(data = point_data, aes(x = x, y = y, fill = z), shape = 21) +
scale_fill_manual(values = c("white", "black"), guide = FALSE) +
labs(x = 'x', y = 'F(x)')
For the second part of your question, you can put all the coordinates in a separate data frame and call geom_point only once:
ddf <- data.frame(xax = rep(c(0:3, 3.5), 2),
cdf = c(0, .5, .6, .8, .9, .5, .6, .8, .9, 1),
col = rep(c("white", "black"), each = 5))
dev.new()
tibble_ex %>%
gather(x, xax, x0:x4) %>%
mutate(cdf = case_when(x == 'x0' ~ 0,
x == 'x1' ~ 1/2,
x == 'x2' ~ 3/5,
x == 'x3' ~ 4/5,
x == 'x35' ~ 9/10,
x == 'x4' ~ 1)) %>%
ggplot(aes(x = xax, y = cdf)) +
geom_step() +
geom_point(data = ddf, aes(fill = I(col)), size = 3, shape = 21) +
labs(x = 'x', y = 'F(x)')
This is my df :
df <- data.frame(annee = rep(c(2003,2004), times = 1, each = 3), sps = c("a", "b", "c"), nb = 1:3)
I create a column containing my labels :
df$labels <- paste("nb", df$sps, "=", df$nb)
Then I do my plot :
ggplot(df, aes(nb)) +
geom_density(aes(fill = sps, colour = sps), alpha = 0.1) +
facet_wrap(~ annee) +
geom_text(data=df, aes(x=8, y=2.5, label= labels), colour="black", inherit.aes=FALSE, parse=FALSE)
But I have a problem with my text in each facet : I would like to have 3 lines (one for each sps).
I tried with the symbol "\n" but I failed in trying to obtain :
"nb a = 1 \n nb b = 2 \n nb c = 3" for each year
Thanks for help
You will have to concatenate what you want broken into several lines into one single string.
newdf <- aggregate(labels ~ annee, data = df, FUN = paste, collapse = "\n")
ggplot(df, aes(nb)) +
geom_density(aes(fill = sps, colour = sps), alpha = 0.1) +
facet_wrap(~ annee) +
geom_text(data = newdf, aes(x = 8, y = 2, label = labels), color = "black") +
scale_x_continuous(limits = c(0, 11)) +
scale_y_continuous(limits = c(0, 2.25))
You can achieve what you want by creating a separate data.frame for your labels:
library(tidyverse)
df <- data.frame(annee = rep(c(2003,2004),
times = 1, each = 3),
sps = c("a", "b", "c"),
nb = 1:3)
# create labels in separate data.frame
label_df <- df %>%
mutate(labels = paste("nb", sps, "=", nb)) %>%
group_by(annee) %>%
summarise(labels = paste(labels, collapse = "\n")) %>%
mutate(x = 6.5,
y = 2.2)
ggplot(df, aes(nb)) +
geom_density(aes(fill = sps, colour = sps), alpha = 0.1) +
facet_wrap(~annee) +
geom_text(data = label_df, aes(x = x, y = y, label = labels)) +
coord_cartesian(ylim = c(0, 2.4), xlim = c(1, 8))
The question relates to this: Line graph customization (add circles, colors), but since I got a new task, I created a new question.
So again my data frame is the same as in the question I've posted in a link. With code below and (little of my own modification) that was given to me by #beetroot
value <- c(9, 4, 10, 7, 10,
10, 10, 4, 10,
4, 10, 2, 5, 5, 4)
names <- c("a","b",
"c","d","e",
"f", "g","h",
"i","j","k","l",
"m","n","p")
df <- data.frame(value, names)
df$names <- as.character(df$names)
df$part <- rep(c("part3", "part2", "part1"), each = 5)
library(dplyr)
library(tidyr)
df2 <- df %>%
group_by(part, names) %>%
expand(value = min(df$value):max(df$value))
p <- ggplot() +
geom_point(data = df2, aes(x = value, y = names),
shape = 1) +
geom_point(data = df, aes(y = names, x = value, group = 1),
colour = I("red"), shape = 21, lwd = 3, fill = "red") +
geom_line(data = df, aes(y = names, x = value, group = 1),
group = I(1),color = I("red")) +
theme_bw() +
facet_wrap(~part, ncol = 1, scales = "free_y")
p + theme(strip.background = element_rect(fill="dodgerblue3"),
strip.text.x = element_text(colour = "white"))+xlab("") +ylab("")
df <- data.frame(value, names)
df$names <- as.character(df$names)
I get this output:
But now I would like to connect lines through (PART1, PART2 and PART3) so that my output would look like:
I used black color of a line just it will be more visible that I would like to connect this parts with lines.
Although I am not completely satisfied I've found solution. I computed the bounding box.
Firstly I removed facet_wrap(~part, ncol = 1, scales = "free_y") so my code looks like this:
p <- ggplot() +
geom_point(data = df2, aes(x = value, y = names),
shape = 1) +
geom_point(data = df, aes(y = names, x = value, group = 1),
colour = I("red"), shape = 21, lwd = 3, fill = "red") +
geom_line(data = df, aes(y = names, x = value, group = 1),
group = I(1),color = I("red")) +
theme_bw()
Then the trick was to create data frame and add the width and height of text directly:
# PART 1
TextFrame <- data.frame(X = 6, Y = 15.5, LAB = "PART 1")
TextFrame <- transform(TextFrame,
w = strwidth(LAB, 'inches') + 8,
h = strheight(LAB, 'inches') + 0.3
)
# PART 2
TextFrame.1 <- data.frame(X = 6, Y = 10.5, LAB = "PART 2")
TextFrame.1 <- transform(TextFrame.1,
w = strwidth(LAB, 'inches') + 8,
h = strheight(LAB, 'inches') + 0.3
)
# PART 3
TextFrame.2 <- data.frame(X = 6, Y = 4.5, LAB = "PART 3")
TextFrame.2 <- transform(TextFrame.2,
w = strwidth(LAB, 'inches') + 8,
h = strheight(LAB, 'inches') + 0.3
)
Then I've used geom_rectand geom_text to create the illusion I am after.
p + geom_rect(data = TextFrame, aes(xmin = X - w/2, xmax = X + w/2,
ymin = Y - h/2, ymax = Y + h/2), fill = "dodgerblue3") +
geom_text(data = TextFrame,aes(x = X, y = Y, label = LAB), size = 5) +
geom_rect(data = TextFrame.1, aes(xmin = X - w/2, xmax = X + w/2,
ymin = Y - h/2, ymax = Y + h/2), fill = "dodgerblue3") +
geom_text(data = TextFrame.1,aes(x = X, y = Y, label = LAB), size = 5) +
geom_rect(data = TextFrame.2, aes(xmin = X - w/2, xmax = X + w/2,
ymin = Y - h/2, ymax = Y + h/2), fill = "dodgerblue3") +
geom_text(data = TextFrame.2,aes(x = X, y = Y, label = LAB), size = 5)
And the output is:
I have a plot from the following script.
require(ggplot2)
df.shape <- data.frame(
AX = runif(10),
AY = runif(10),
BX = runif(10, 2, 3),
BY = runif(10, 2, 3)
)
p <- ggplot(df.shape)
p <- p + geom_point(aes(x = AX, y = AY, shape = 15)) +
geom_point(aes(x = BX, y = BY, shape = 19)) +
scale_shape_identity() +
guides(shape = guide_legend(override.aes = list(shape = 15, shape = 19)) )
print(p)
This doesn't produce a legend, describing which shape is "A" and which shape is "B". Note that the squares and circles may be close to one another, so I can't generally define the variable based on location. How do I display a "shape" legend?
I would reshape my data in the long format using reshape:
dt <- reshape(df.shape ,direction='long', varying=list(c(1, 3), c(2, 4)),
,v.names = c('X','Y'), times = c('A','B'))
Then I plot it simply like this
ggplot(dt) +
geom_point(aes(x = X, y = Y, shape = time),size=5) +
scale_shape_manual(values=c(15,19))