Plot Curved Lines Between Two Locations in Ggplot2

Plot curved lines between two locations in ggplot2

I couldn't run cshapes for some reason, but here's an example of how to build curves using curveGrob() from the grid package and ggplot2's annotation_custom() function. It gives you a lot of flexibility. PS: most of the params are just defaults. Edit - updated to show 2 curves.

require(grid)
g<-qplot(c(0,10),c(0,10))
myCurve<-curveGrob(0, 0, 1, 1, default.units = "npc",
curvature = 0.3, angle = 90, ncp = 20, shape = 1,
square = FALSE, squareShape = 1,
inflect = FALSE, arrow = arrow(), open = TRUE,
debug = FALSE,
name = NULL, gp = gpar(), vp = NULL)

myCurve2<-curveGrob(0, 0, 1, 1, default.units = "npc",
curvature = -0.3, angle = 60, ncp = 10, shape = 1,
square = FALSE, squareShape = 1,
inflect = FALSE, arrow = arrow(), open = TRUE,
debug = FALSE,
name = NULL, gp = gpar(), vp = NULL)

g +
annotation_custom(grob=myCurve,0,10,0,10) + # plot from 0,0 to 10,10
annotation_custom(grob=myCurve2,2.5,6,2.5,6) # plot from 2.5,2.5 to 6,6

#REFERENCE>>http://stat.ethz.ch/R-manual/R-devel/library/grid/html/grid.curve.html

Sample Image

How to plot a curve line between points

One way get smoothed lines instead of straight lines would be to flip x and y in your aesthetics, then apply geom_smooth instead of geom_path and then flip the coordinates through coord_flip:

ggplot(tab, aes(x=Sample, y=N, c(0,0.16)),pch=17) + 
coord_flip() +
geom_point(color='#2980B9', size = 2) +
geom_smooth(method = "loess", se = FALSE,
span = 0.25, linetype=3,color='#2980B9', size = 0.1)

Sample Image

connecting points with curved line on ggplot for a categorical variable on the 'x' axis

Something like this?

library(dplyr)

ggplot(oi2, aes(x = trial, y = prop, color = rs.)) +
geom_point() +
geom_line(data = . %>%
mutate(trial = as.numeric(trial)) %>%
group_by(rs.) %>%

# increase n for smoother line; can also try the other methods listed under
# ?spline, though I find "natural" looks better than some of the rest: "fmm"'s
# curves are rather drastic, & "periodic" doesn't touch all the points
summarise(x1 = list(spline(trial, prop, n = 50, method = "natural")[["x"]]),
y1 = list(spline(trial, prop, n = 50, method = "natural")[["y"]])) %>%

tidyr::unnest(),
aes(x = x1, y = y1)) +
labs(subtitle="Favorable allele over time",
y="allele frequency",
x="Groups",
title="Yield QTL markers",
col='markers')

plot

Off topic: There are probably too many unique markers in the chart for colour to serve as an effective way to distinguish between them. You may be better off splitting this into multiple charts, with fewer markers in each. Or only give distinctive colours to the ones you wish to highlight, & leave the rest grey.

plot (ggplot ?) smooth + color area between 2 curves

Cool question since I had to give myself a crash course in using LOESS for ribbons!

First thing I'm doing is getting the data into a long shape, since that's what ggplot will expect, and since your data has some characteristics that are kind of hidden within values. For example, if you gather into a long shape and have, say a column key, with a value of "inf20" and another of "sup20", those hold more information than you currently have access to, i.e. the measure type is either "inf" or "sup", and the level is 20. You can extract that information out of that column to get columns of measure types ("inf" or "sup") and levels (20, 40, 60, or 90), then map aesthetics onto those variables.

So here I'm getting the data into a long shape, then using spread to make columns of inf and sup, because those will become ymin and ymax for the ribbons. I made level a factor and reversed its levels, because I wanted to change the order of the ribbons being drawn such that the narrow one would come up last and be drawn on top.

library(tidyverse)

data_long <- data %>%
as_tibble() %>%
gather(key = key, value = value, -Nb_obs, -Nb_obst) %>%
mutate(measure = str_extract(key, "\\D+")) %>%
mutate(level = str_extract(key, "\\d+")) %>%
select(-key) %>%
group_by(level, measure) %>%
mutate(row = row_number()) %>%
spread(key = measure, value = value) %>%
ungroup() %>%
mutate(level = as.factor(level) %>% fct_rev())

head(data_long)
#> # A tibble: 6 x 6
#> Nb_obs Nb_obst level row inf sup
#> <dbl> <dbl> <fct> <int> <dbl> <dbl>
#> 1 0 35 20 2 2 4
#> 2 0 35 40 2 2 5
#> 3 0 35 60 2 1 6
#> 4 0 35 90 2 0 11
#> 5 0 39 20 8 3 5
#> 6 0 39 40 8 2 6

ggplot(data_long, aes(x = Nb_obst, ymin = inf, ymax = sup, fill = level)) +
geom_ribbon(alpha = 0.6) +
scale_fill_manual(values = c("20" = "darkred", "40" = "red",
"60" = "darkorange", "90" = "yellow")) +
theme_light()

Sample Image

But it still has the issue of being jagged, so for each level I predicted smoothed values of both inf and sup versus Nb_obst using loess. group_by and do yield a nested data frame, and unnest pulls it back out into a workable form. Feel free to adjust the span parameter, as well as other loess.control parameters that I know very little about.

data_smooth <- data_long %>%
group_by(level) %>%
do(Nb_obst = .$Nb_obst,
inf_smooth = predict(loess(.$inf ~ .$Nb_obst, span = 0.35), .$Nb_obst),
sup_smooth = predict(loess(.$sup ~ .$Nb_obst, span = 0.35), .$Nb_obst)) %>%
unnest()

head(data_smooth)
#> # A tibble: 6 x 4
#> level Nb_obst inf_smooth sup_smooth
#> <fct> <dbl> <dbl> <dbl>
#> 1 90 35 0 11.
#> 2 90 39 0 13.4
#> 3 90 48 0.526 16.7
#> 4 90 39 0 13.4
#> 5 90 41 0 13
#> 6 90 41 0 13

ggplot(data_smooth, aes(x = Nb_obst, ymin = inf_smooth, ymax = sup_smooth, fill = level)) +
geom_ribbon(alpha = 0.6) +
scale_fill_manual(values = c("20" = "darkred", "40" = "red",
"60" = "darkorange", "90" = "yellow")) +
theme_light()

Sample Image

Created on 2018-05-26 by the reprex package (v0.2.0).

R ggplot multiple series curved line

As @MrFlick mentions in the comments, there are serious statistical ways of getting curved lines, which are probably off topic here.

If you just want your graph to look nicer however, you could try interpolating your data with spline, then adding it on as another layer.

First we make some spline data, using 10 times the number of data points you had (you can increase or decrease this as desired):

library(dplyr)
dat2 <- td_results %>% select(count_id, AMV, duration_in_traffic) %>%
group_by(count_id) %>%
do(as.data.frame(spline(x= .[["AMV"]], y= .[["duration_in_traffic"]], n = nrow(.)*10)))

Then we plot, using your original data for points, but then using lines from the spline data (dat2):

library(ggplot2)
ggplot(td_results, aes(AMV, duration_in_traffic)) +
geom_point(aes(colour = factor(count_id))) +
geom_line(data = dat2, aes(x = x, y = y, colour = factor(count_id)))

This gives me the following graph from your test data:
Sample Image

Connecting two points with curved lines (s-ish curve) in R

Following @thelatemail's suggestion, I decided to make my edit into an answer. My solution is based on @thelatemail's answer.

I wrote a small function to draw curves, which makes use of the logistic function:

#Create the function
curveMaker <- function(x1, y1, x2, y2, ...){
curve( plogis( x, scale = 0.08, loc = (x1 + x2) /2 ) * (y2-y1) + y1,
x1, x2, add = TRUE, ...)
}

A working example is below. In this example, I want to create a plot for a taxonomy with 3 levels: parent --> 2 children -- > 20 grandchildren. One child has 12 grandchildren, and the other child has 8 children.

#Prepare data:
parent <- c(1, 16)
children <- cbind(2, c(8, 28))
grandchildren <- cbind(3, (1:20)*2-1)
labels <- c("Parent ", paste("Child ", 1:2), paste(" Grandchild", 1:20) )

#Make a blank plot canvas
plot(0, type="n", ann = FALSE, xlim = c( 0.5, 3.5 ), ylim = c( 0.5, 39.5 ), axes = FALSE )

#Plot curves
#Parent and children
invisible( mapply( curveMaker,
x1 = parent[ 1 ],
y1 = parent[ 2 ],
x2 = children[ , 1 ],
y2 = children[ , 2 ],
col = gray( 0.6, alpha = 0.6 ), lwd = 1.5 ) )

#Children and grandchildren
invisible( mapply( curveMaker,
x1 = children[ 1, 1 ],
y1 = children[ 1, 2 ],
x2 = grandchildren[ 1:8 , 1 ],
y2 = grandchildren[ 1:8, 2 ],
col = gray( 0.6, alpha = 0.6 ), lwd = 1.5 ) )
invisible( mapply( curveMaker,
x1 = children[ 2, 1 ],
y1 = children[ 2, 2 ],
x2 = grandchildren[ 9:20 , 1 ],
y2 = grandchildren[ 9:20, 2 ],
col = gray( 0.6, alpha = 0.6 ), lwd = 1.5 ) )
#Plot text
text( x = c(parent[1], children[,1], grandchildren[,1]),
y = c(parent[2], children[,2], grandchildren[,2]),
labels = labels,
pos = rep(c(2, 4), c(3, 20) ) )

#Plot points
points( x = c(parent[1], children[,1], grandchildren[,1]),
y = c(parent[2], children[,2], grandchildren[,2]),
pch = 21, bg = "white", col="#3182bd", lwd=2.5, cex=1)

Sample Image

Draw curved lines in ggmap, geom_curve not working

I think using coord_cartesian() give you what you want. Using the 5 rows showed in your post

map <- get_map(location = "Amsterdam", zoom = 11)
ggmap(map) +
geom_point(data = df_vertices,
aes(x = Longitude, y = Latitude, size = scaledkWh),
colour = "red", alpha =0.5) +
geom_curve(data = df,
aes(x = Longitude_from, y = Latitude_from, xend = Longitude_to, yend = Latitude_to),
arrow = arrow(angle = 15, ends = "first", length = unit(0.5, "cm"), type = "closed"),
size = df$scaledAmount, alpha = 0.5, curvature = 0.15, inherit.aes = TRUE)
scale_size_continuous(range=c(1,30)) +
coord_cartesian()

Sample Image



Related Topics



Leave a reply



Submit