I don't know what am I missing in the code?
set.seed(12345)
require(ggplot2)
AData <- data.frame(Glabel=LETTERS[1:7], A=rnorm(7, mean = 0, sd = 1), B=rnorm(7, mean = 0, sd = 1))
TData <- data.frame(Tlabel=LETTERS[11:20], A=rnorm(10, mean = 0, sd = 1), B=rnorm(10, mean = 0, sd = 1))
i <- 2
j <- 3
p <- ggplot(data=AData, aes(AData[, i], AData[, j])) + geom_point() + theme_bw()
p <- p + geom_text(aes(data=AData, label=Glabel), size=3, vjust=1.25, colour="black")
p <- p + geom_segment(data = TData, aes(xend = TData[ ,i], yend=TData[ ,j]),
x=0, y=0, colour="black",
arrow=arrow(angle=25, length=unit(0.25, "cm")))
p <- p + geom_text(data=TData, aes(label=Tlabel), size=3, vjust=1.35, colour="black")
Last line of the code produces the error. Please point me out how to figure out this problem. Thanks in advance.
I have no idea what you are trying to do, but the line that fails is the last line, because you haven't mapped new x and y variables in the mapping. geom_text() needs x and y coords but you only provide the label argument, so ggplot takes x and y from p, which has only 7 rows of data whilst Tlabel is of length 10. That explains the error. I presume you mean to plot at x = A and y = B of TData? If so, this works:
p + geom_text(data=TData, mapping = aes(A, B, label=Tlabel),
size=3, vjust=1.35, colour="black")
(This might get a better answer on the ggplot mailing list.)
It looks like you're trying to display some kind of biplot ... the root of your problem is that you're violating the idiom of ggplot, which wants you to specify variables in a way that's consistent with the scope of the data.
Maybe this does what you want, via some aes_string trickery that substitutes the names of the desired columns ...
varnames <- colnames(AData)[-1]
v1 <- varnames[1]
v2 <- varnames[2]
p <- ggplot(data=AData,
aes_string(x=v1, y=v2)) + geom_point() + theme_bw()
## took out redundant 'data', made size bigger so I could see the labels
p <- p + geom_text(aes(label=Glabel), size=7, vjust=1.25, colour="black")
p <- p + geom_segment(data = TData, aes_string(xend = v1, yend=v2),
x=0, y=0, colour="black",
arrow=arrow(angle=25, length=unit(0.25, "cm")))
## added colour so I could distinguish this second set of labels
p <- p + geom_text(data=TData,
aes(label=Tlabel), size=10, vjust=1.35, colour="blue")
Related
Is there any way to set the break step size in ggplot without defining a sequence. For example:
x <- 1:10
y <- 1:10
df <- data.frame(x, y)
# Plot with auto scale
ggplot(df, aes(x,y)) + geom_point()
# Plot with breaks defined by sequence
ggplot(df, aes(x,y)) + geom_point() +
scale_y_continuous(breaks = seq(0,10,1))
# Plot with automatic sequence for breaks
ggplot(df, aes(x,y)) + geom_point() +
scale_y_continuous(breaks = seq(min(df$y),max(df$y),1))
# Does this exist?
ggplot(df, aes(x,y)) + geom_point() +
scale_y_continuous(break_step = 1)
You may say I am being lazy but there have been a few occasions where I have had to change the min and max limits of my seq due to the addition of error bars. So I just want to say...use a break size of x, with automatic scale limits.
You can define your own function to pass to the breaks argument. An example that would work in your case would be
f <- function(y) seq(floor(min(y)), ceiling(max(y)))
Then
ggplot(df, aes(x,y)) + geom_point() + scale_y_continuous(breaks = f)
gives
You could modify this to pass the step of the breaks, e.g.
f <- function(k) {
step <- k
function(y) seq(floor(min(y)), ceiling(max(y)), by = step)
}
then
ggplot(df, aes(x,y)) + geom_point() + scale_y_continuous(breaks = f(2))
would create a y-axis with ticks at 2, 4, .., 10, etc.
You can take this even further by writing your own scale function
my_scale <- function(step = 1, ...) scale_y_continuous(breaks = f(step), ...)
and just call it like
ggplot(df, aes(x,y)) + geom_point() + my_scale()
> # Does this exist?
> ggplot(df, aes(x,y)) + geom_point() +
> scale_y_continuous(break_step = 1)
If you're looking for an off-the-shelf solution, then you can use the scales::breaks_width() function like so:
scale_y_continuous(breaks = scales::breaks_width(1))
The scales package also includes handy functions to control breaks easily in "special" scales such as date-time, e.g. scale_x_datetime(breaks='6 hours').
I am using ggplot geom_smooth to plot turnover data of a customer group from previous year against the current year (based on calendar weeks). As the last week is not complete, I would like to use a dashed linetype for the last week. However, I can't figure out how to that. I can either change the linetype for the entire plot or an entire series, but not within a series (depending on the value of x):
To keep it simple, let's just use the following example:
set.seed(42)
frame <- data.frame(series = rep(c('a','b'),50),x = 1:100, y = runif(100))
ggplot(frame,aes(x = x,y = y, group = series, color=series)) +
geom_smooth(size=1.5, se=FALSE)
How would I have to change this to get dashed lines for x >= 75?
The goal would be something like this:
Thx very much for any help!
Edit, 2016-03-05
Of course I fail when trying to use this method on the original plot. The Problem lies with the ribbon, which is calculated using stat_summary and a predefined function. I tried to use use stat_summary on the original data (mdf), and geom_line on the smooth_data. Even when I comment out everything else, I still get "Error: Continuous value supplied to discrete scale". I believe the problem comes from the fact that the original x value (Kalenderwoche) was discrete, whereas the new, smoothed x is continuous. Do I have to somehow transform one into the other? What else could I do?
Here is what I tried (condensed to the essential lines):
quartiles <- function(x) {
x <- na.omit(x) # remove NULL
median <- median(x)
q1 <- quantile(x,0.25)
q3 <- quantile(x,0.75)
data.frame(y = median, ymin = median, ymax = q3)
}
g <- ggplot(mdf, aes(x=Kalenderwoche, y=value, group=variable, colour=variable,fill=variable))+
geom_smooth(size=1.5, method="auto", se=FALSE)
# Take out the data for smooth line
smooth_data <- ggplot_build(g)$data[[1]]
ggplot(mdf, aes(x=Kalenderwoche, y=value, group=variable, colour=variable,fill=variable))+
stat_summary(fun.data = quartiles,geom="ribbon", colour="NA", alpha=0.25)+
geom_line(data=smooth_data, aes(x=x, y=y, group=group, colour=group, fill=group))
mdf looks like this:
str(mdf)
'data.frame': 280086 obs. of 5 variables:
$ konto_id : int 1 1 1 1 1 1 1 1 1 1 ...
$ Kalenderwoche: Factor w/ 14 levels "2015-48","2015-49",..: 4 12 1 3 7 13 10 6 5 9 ...
$ variable : Factor w/ 2 levels "Umsatz","Umsatz Vorjahr": 1 1 1 1 1 1 1 1 1 1 ...
$ value : num 0 428.3 97.8 76 793.1 ...
There are many accounts (konto_id), and for each account and calendar week (Kalenderwoche), there is a current turnover value (Umsatz) and a turnover value from last year (Umsatz Vorjahr). I can provide a smaller version of the data.frame and the entire code, if required.
Thx very much for any help!
P.S. I am a total novice in R, so my code probably looks rather stupid to pros, sorry for that :(
Edit, 2016-03-06
I have uploaded a subset of the data (mdf):
mdf
The full code of the original graph is the following (looking somewhat weird with so little data, but that's not the point ;)
library(dtw)
library(reshape2)
library(ggplot2)
library(RODBC)
library(Cairo)
# custom breaks for X axis
breaks.custom <- unique(mdf$Kalenderwoche)[c(TRUE,rep(FALSE,0))]
# function called by stat_summary
quartiles <- function(x) {
x <- na.omit(x)
median <- median(x)
q1 <- quantile(x,0.25)
q3 <- quantile(x,0.75)
data.frame(y = median, ymin = median, ymax = q3)
}
# Positions for guidelines and labels
horizontal.center <- (length(unique(mdf$Kalenderwoche))+1)/2
kw.horizontal.center <- as.vector(sort(unique(mdf$Kalenderwoche))[c(horizontal.center-0.5,horizontal.center+0.5)])
vpos.P75.label <- max(quantile(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[1]],0.75)
,quantile(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[2]],0.75))+10
# use the higher P75 value of the two weeks around the center
vpos.mean.label <- min(mean(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[1]])
,mean(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[2]]))-10
vpos.median.label <- min(median(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[1]])
,median(mdf$value[mdf$Kalenderwoche==kw.horizontal.center[2]]))-10
hpos.vline <- which(as.vector(sort(unique(mdf$Kalenderwoche))=="2016-03"))
# custom colour palette (2 colors)
cbPaletteLine <- c("#DA2626", "#2626DA")
cbPaletteFill <- c("#F0A8A8", "#7C7CE9")
# ggplot
ggplot(mdf, aes(x=Kalenderwoche, y=value, group=variable, colour=variable,fill=variable))+
geom_smooth(size=1.5, method="auto", se=FALSE)+
# SE=FALSE to suppress drawing of the SE of the fit.SE of the data shall be used instead:
stat_summary(fun.data = quartiles,geom="ribbon", colour="NA", alpha=0.25)+
scale_x_discrete(breaks=breaks.custom)+
scale_colour_manual(values=cbPaletteLine)+
scale_fill_manual(values=cbPaletteFill)+
#coord_cartesian(ylim = c(0, 250)) +
theme(legend.title = element_blank(), title = element_text(face="bold", size=12))+
#scale_color_brewer(palette="Dark2")+
labs(title = "Tranche 1", x = "Kalenderwoche", y = "Konto-Umsatz [CHF]")+
geom_vline(xintercept = hpos.vline, linetype=2)+
annotate("text", x=horizontal.center, y=vpos.median.label, label = "Median", size=4)+
annotate("text", x=horizontal.center, y=vpos.mean.label, label= "Mean", size=4)+
annotate("text", x=horizontal.center, y=vpos.P75.label, label = "P75%", size=4)+
theme(axis.text.x=element_text(angle = 90, hjust = 0.5, vjust = 0.5))
Edit, 2016-03-06
The final plot now looks like this (thx, Jason!!)
I am not so sure how to smooth all data and use different line types for subsets by geom_smooth function. My idea is to pull out the data which ggplot used to construct the plot and use geom_line to reproduce it. This was the way I did it:
set.seed(42)
frame <- data.frame(series=rep(c('a','b'), 50),
x = 1:100, y = runif(100))
library(ggplot2)
g <- ggplot(frame, aes(x=x, y=y, color=series)) + geom_smooth(se=FALSE)
# Take out the data for smooth line
smooth_data <- ggplot_build(g)$data[[1]]
ggplot(smooth_data[smooth_data$x <= 76, ], aes(x=x, y=y, color=as.factor(group), group=group)) +
geom_line(size=1.5) +
geom_line(data=smooth_data[smooth_data$x >= 74, ], linetype="dashed", size=1.5) +
scale_color_discrete("Series", breaks=c("1", "2"), labels=c("a", "b"))
You're right. The problem is that you add a continuous x to a discrete x in the original layer. One way to deal with it is to create a lookup table which in this case, it is easy because x is a sequence from 1 to 14. We can transform discrete x by indexing. In your code, it should work if you add:
level <- levels(mdf$Kalenderwoche)
ggplot(mdf, aes(x=Kalenderwoche, y=value, group=variable, colour=variable,fill=variable))+
stat_summary(fun.data = quartiles,geom="ribbon", colour="NA", alpha=0.25) +
geom_line(data=smooth_data, aes(x=level[x], y=y, group=group, colour=as.factor(group), fill=NA))
Here is my attempt for the question:
g <- ggplot(mdf, aes(x=Kalenderwoche, y=value, group=variable, colour=variable,fill=variable)) +
geom_smooth(size=1.5, method="auto", se=FALSE) +
# SE=FALSE to suppress drawing of the SE of the fit.SE of the data shall be used instead:
stat_summary(fun.data = quartiles,geom="ribbon", colour="NA", alpha=0.25)
smooth_data <- ggplot_build(g)$data[[1]]
ribbon_data <- ggplot_build(g)$data[[2]]
# Use them as lookup table
level <- levels(mdf$Kalenderwoche)
clevel <- levels(mdf$variable)
ggplot(smooth_data[smooth_data$x <= 13, ], aes(x=level[x], y=y, group=group, color=as.factor(clevel[group]))) +
geom_line(size=1.5) +
geom_line(data=smooth_data[smooth_data$x >= 13, ], linetype="dashed", size=1.5) +
geom_ribbon(data=ribbon_data,
aes(x=x, ymin=ymin, ymax=ymax, fill=as.factor(clevel[group]), color=NA), alpha=0.25) +
scale_x_discrete(breaks=breaks.custom) +
scale_colour_manual(values=cbPaletteLine) +
scale_fill_manual(values=cbPaletteFill) +
#coord_cartesian(ylim = c(0, 250)) +
theme(legend.title = element_blank(), title = element_text(face="bold", size=12))+
#scale_color_brewer(palette="Dark2")+
labs(title = "Tranche 1", x = "Kalenderwoche", y = "Konto-Umsatz [CHF]")+
geom_vline(xintercept = hpos.vline, linetype=2)+
annotate("text", x=horizontal.center, y=vpos.median.label, label = "Median", size=4)+
annotate("text", x=horizontal.center, y=vpos.mean.label, label= "Mean", size=4)+
annotate("text", x=horizontal.center, y=vpos.P75.label, label = "P75%", size=4)+
theme(axis.text.x=element_text(angle = 90, hjust = 0.5, vjust = 0.5))
Note that the legend has borderline.
I know there have been a number of entries with regards to adding R^2 values to plots, but I am having trouble following the codes. I am graphing a scatter plot with three categories. I have added a linear regression line for each one. I would now like to add r^2 values for each but I can't figure out how to do this.
My code:
veg <- read.csv("latandwtall2.csv", header=TRUE)
library("ggplot2")
a <- ggplot(veg, aes(x=avglat, y=wtfi, color=genus)) + geom_point(shape=19, size=4)
b <- a + scale_colour_hue(l=50) + stat_smooth(method = "lm", formula = y ~ x, size = 1, se = FALSE)
c <- b + labs(x="Latitude", y="Weight (g)")
d <- c + theme_bw()
e <- d + theme(panel.grid.minor=element_blank(), panel.grid.major=element_blank())
#changes size of text
f <- e + theme(
axis.title.x = element_text(color="black", vjust=-0.35, size=15, face="bold"),
axis.title.y = element_text(color="black" , vjust=0.35, size=15, face="bold")
)
g <- e+theme(legend.key=element_rect(fill='white'))
g
Any help with how to add R^2 values would be greatly appreciated. Thanks!
If you build a data frame with the r-squared values, you might be able to (mostly) automate the positioning of the annotation text by including it as a call to geom_text.
Here's a toy example. The rsq data frame is used in geom_text to place the r-squared labels. In this case, I've set it up to put the labels just after the highest x-value and the predict function gets the y-value. It's probably too much work for a single plot, but if you're doing this a lot, you can turn it into a function so that you don't have to repeat the set-up code every time, and maybe add some fancier logic to make label placement more flexible:
library(reshape2) # For melt function
# Fake data
set.seed(12)
x = runif(100, 0, 10)
dat = data.frame(x, y1 = 2*x + 3 + rnorm(100, 0, 5),
y2 = 4*x + 20 + rnorm(100, 0, 10))
dat.m = melt(dat, id.var="x")
# linear models
my1 = lm(y1 ~ x, data=dat)
my2 = lm(y2 ~ x, data=dat)
# Data frame for adding r-squared values to plot
rsq = data.frame(model=c("y1","y2"),
r2=c(summary(my1)$adj.r.squared,
summary(my2)$adj.r.squared),
x=max(dat$x),
y=c(predict(my1, newdata=data.frame(x=max(dat$x))),
predict(my2, newdata=data.frame(x=max(dat$x)))))
ggplot() +
geom_point(data=dat.m, aes(x, value, colour=variable)) +
geom_smooth(data=dat.m, aes(x, value, colour=variable),
method="lm", se=FALSE) +
geom_text(data=rsq, aes(label=paste("r^2 == ", round(r2,2)),
x=1.05*x, y=y, colour=model, hjust=0.5),
size=4.5, parse=TRUE)
I can't really reproduce what you're doing but you need to use annotate()
Something that could work (puting the R2 on the 10th point) would be :
R2 = 0.4
i = 10
text = paste("R-squared = ", R2, sep="")
g = g + annotate("text", x=avglat[i], y=wtfi[i], label=text, font="Calibri", colour="red", vjust = -2, hjust = 1)
Use vjust and hjust to adjust the position of the text to the point (change the i), and just fill the variable R2 with your computed rsquared. You can choose the point you like or manually enter the x,y coordinate it's up to you. Does that help ?
PS. I put extra parameters (font, colours) so that you have the flexibility to change them.
Build the model separately, get the R^2 from there, and add it to the plot. I'll give you some dummy code, but it would be of better quality if you had given us a sample data frame.
r2 = summary(lm(wtfi ~ avglat, data=veg))$r.squared
#to piggyback on Romain's code...
i=10
g = g + annotate("text", x=avglat[i], y=wtfi[i], label=round(r2,2), font="Calibri", colour="red", vjust = -2, hjust = 1)
The way I wrote it here you don't need to hard-code the R^2 value in.
I have following code to draw my logistic distribution:
x=seq(-2000,2000,length=1000)
dat <- data.frame(x=x)
dat$value <- dlogis(x,location=200,scale=400/log(10))
dat$type <- "Expected score"
p <- ggplot(data=dat, aes(x=x, y=value)) + geom_line(col="blue", size=1) +
coord_cartesian(xlim = c(-500, 900), ylim = c(0, 0.0016)) +
scale_x_continuous(breaks=c(seq(-500, 800, 100)))
pp <- p + geom_line(aes(x = c(0,0), y = c(0,0.0011)), size=0.9, colour="green", linetype=2, alpha=0.7)
Now what I would like to do is to highlight the area to the left of x = 0.
I tried to do it like this:
x = seq(-500, 0, length=10)
y = dlogis(x,location=200,scale=400/log(10))
pol <- data.frame(x = x, y = y)
pp + geom_polygon(aes(data=pol,x=x, y=y), fill="light blue", alpha=0.6)
But this does not work. Not sure what I am doing wrong. Any help?
I haven't diagnosed the problem with your polygon (although I think you would need to give the full path around the outside, i.e. attach rep(0,length(x)) to the end of y and rev(x) to the end of x), but geom_ribbon (as in Shading a kernel density plot between two points. ) seems to do the trick:
pp + geom_ribbon(data=data.frame(x=x,y=y),aes(ymax=y,x=x,y=NULL),
ymin=0,fill="light blue",alpha=0.5)
UPDATE:
I have solved my problem. I was looking for
coord_cartesian(xlim = c(800, 2100), ylim = c(0, 0.0021))
Thanks to every one who tried to help!
QUESTION WAS:
I would like to draw a nice picture of what is the difference between normal and logistic distributions. I have reached that point :
x=seq(1000,2000,length=200)
dat <- data.frame(
norm = dnorm(x,mean=1500,sd=200),
logistic = dlogis(x,location=1500,scale=200), x = x
)
ggplot(data=dat, aes(x=x)) +
geom_polygon(aes(y=norm), fill="red", alpha=0.6) +
geom_polygon(aes(y=logistic), fill="blue", alpha=0.6) +
xlab("") + ylab("") +
opts(title="Logistic and Normal Distributions") +
scale_x_continuous(expand = c(0, 0)) +
scale_y_continuous(expand = c(0, 0))
However the logistic one is "cut" at the bottom. I think what I should do is to draw this distribution from 0 to 3000 for example but show only 1000-2000.
Any clues how to do this?
I tried scale_x_continuous(limits = c(1000, 2000)) but this does not work
UPDATE:
I have updated my code so I have legend, now it looks like this:
x=seq(700,2300,length=200)
dat2 <- data.frame(x=x)
dat2$value <- dnorm(x,mean=1500,sd=200)
dat2$type <- "Normal"
dat1 <- data.frame(x=x)
dat1$value <- dlogis(x,location=1500,scale=200)
dat1$type <- "Logistic"
dat <- rbind(dat1, dat2)
ggplot(data=dat, aes(x=x, y=value, colour=type, fill=type)) + geom_polygon(alpha=0.6) + scale_y_continuous(expand = c(0, 0))
I would draw it using z-scores, from [-2 ; +2]. This has the side benefit that your problem goes away.
x=seq(-2,2,length=200)
dat <- data.frame(
norm = dnorm(x,mean=0,sd=0.2),
logistic = dlogis(x,location=0,scale=0.2), x = x
)
p <- ggplot(data=dat, aes(x=x)) +
geom_polygon(aes(y=norm), fill="red", alpha=0.6) +
geom_polygon(aes(y=logistic), fill="blue", alpha=0.6) +
xlab("z") + ylab("") +
scale_x_continuous(expand = c(0, 0)) +
scale_y_continuous(expand = c(0, 0)) +
opts(title="Logistic and Normal Distributions")
print(p)
The reason it cuts off the bottom is because geom_polygon literally draws the polygon consisting of lines connecting the points you give it. So the flat line across the bottom of the distribution is just connecting the first and last value in your data frame. If you want it to extend to the bottom you can add the appropriate points to your data frame:
ggplot(data=dat, aes(x=x)) +
geom_polygon(aes(y=norm), fill="red", alpha=0.6) +
geom_polygon(data = rbind(c(NA,0,1000),dat,c(NA,0,2000)),aes(y=logistic), fill="blue", alpha=0.6) + xlab("") + ylab("") +
opts(title="Logistic and Normal Distributions")+
scale_x_continuous(expand = c(0, 0)) +
scale_y_continuous(expand = c(0, 0))
Edited for clarity
You can tinker with this to get it to go down only as far as you want by adding points with the right values. For instance, I forced the logistic distribution to fill all the way down to zero. You could make it level with the normal distribution by rbinding the minimum normal density value instead. Also, be careful where you add them in your data frame. geom_polygon will connect the dots in the order they appear. That's why I added one at the beginning of the data frame and one at the end.
Edit 2
Based on your revised code, my solution still works fine:
x=seq(700,2300,length=200)
dat2 <- data.frame(x=x)
dat2$value <- dnorm(x,mean=1500,sd=200)
dat2$type <- "Normal"
dat1 <- data.frame(x=x)
dat1$value <- dlogis(x,location=1500,scale=200)
dat1$type <- "Logistic"
#Append extra points at the top/bottom to
# complete the polygon
dat1 <- rbind(data.frame(x=700,value=0,type = "Logistic"),dat1,
data.frame(x=2300,value=0,type = "Logistic"))
dat <- rbind(dat1, dat2)
ggplot(data=dat, aes(x=x, y=value, colour=type, fill=type)) +
geom_polygon(alpha=0.6) +
scale_y_continuous(expand = c(0, 0))
And personally, I would prefer this over coord_cartesian, since I'm a stickler about starting my axes from zero.
The solution is to use
+ coord_cartesian(xlim = c(800, 2100), ylim = c(0, 0.0021))
I ran your code, and then analyzed the values of norm and logistic:
Rgames: mystat(dat$logistic)
min max mean median
3.51e-04 1.25e-03 8.46e-04 8.63e-04
sdev skew kurtosis
2.96e-04 -1.33e-01 -1.4
Rgames: mystat(dat$norm)
min max mean median
8.76e-05 1.99e-03 9.83e-04 9.06e-04
sdev skew kurtosis
6.62e-04 1.67e-01 -1.48
So your logistic values are in fact correctly plotted. As the other answers showed, there are preferable ways to create your underlying data.