CODEWITHSUNDEEP
r
Duna
Duna

Reputation: 735

How can I adjust the axes to start from zero origin in r plot

To plot the empirical cumulative density of three variables, x1, x2 and x3, I used the following in r: 

plot.ecdf(x1, col="blue",
     main="Distribution XYZ",
     xlab="x_i", ylab="Prob(x_i<=y)")

lines(ecdf(x2), col="red")    # adds a line
lines(ecdf(x3), col="green")  # adds line
legend(600,0.6, # places a legend from (x,y)=(600,0.6) on
       c("x1","x2","x3"), # puts text in the legend
       lty=c(1,1,1), # gives the legend appropriate symbols (lines)
       lwd=c(1,1,1),col=c("blue","red","green")) # gives the legend lines the correct color and width

The resulting plot however has two horizontal lines (broken lines) at 0 and 1 besides the box. And, the origin of the box has a space below zero on the vertical axis and a space to the left of zero on the horizontal axis. May you suggest how to remove these space and the additional lines. I wanted but could not post the plot.

EDITED: A sample data can be generated as follows:

sample data

n <- 1000; u <- runif(n)
a <- -4.46; b <- 1.6; c <- -4.63
d <- ( a * u ) + (b * ( ( 1.5 * ( u ** 2 )) - 0.5 ))  + (c * ( (2.5 * (u ** 3)) - (1.5 * u )))
x1 <- -126/d; x2 <- -131/d; x3 <- -187/d

Upvotes: 17

Views: 49063

Answers (3)

IRTFM
IRTFM

Reputation: 263301

It sounds like you are asking for the style provided by different settings for 'xaxs' and 'yaxs', and maybe 'xlim':

Try:

 plot.ecdf(x1, col="blue",
     main="Distribution XYZ",
     xlab="x_i", ylab="Prob(x_i<=y)", 
     xaxs="i",yaxs="i", xlim=c(0,1000) )

lines(ecdf(x2), col="red")    
lines(ecdf(x3), col="green")  
legend(600,0.6, 
       c("x1","x2","x3"), 
       lty=c(1,1,1), 
       lwd=c(1,1,1),col=c("blue","red","green"))

enter image description here

Upvotes: 17

Henrik
Henrik

Reputation: 14450

use the yaxs = "i" argument:

plot.ecdf(x1, col="blue",
     main="Distribution XYZ",
     xlab="x_i", ylab="Prob(x_i<=y)", yaxs = "i")

This will align the axis at the ylim points (which supposedly are 0 and 1 for plot.ecdf).

If you also want to remove the dotted line at 0 and 1, just call box():

box()

Upvotes: 9

user1981275
user1981275

Reputation: 13372

You could use the arguments xlim, ylim and col.01line: 

x1 = runif(100)
x2 = runif(100)
x3 = runif(100)
plot.ecdf(x1, col="blue", main="Distribution XYZ",xlab="x_i", ylab="Prob(x_i<=y)", ylim=c(0, 1), xlim=c(0,1), col.01line="white", verticals=FALSE)
lines(ecdf(x2), col="red", col.01line="white")
lines(ecdf(x3), col="green", col.01line="white")
legend(600,0.6,c("x1","x2","x3"), lty=c(1,1,1), lwd=c(1,1,1),col=c("blue","red","green"))

enter image description here

Alternatively, you could just use the generic plot:

f1 = ecdf(x1)
f2 = ecdf(x2)
f3 = ecdf(x3)
pts = seq(0, 1, 0.01)
plot(pts, f1(pts), col="blue", type="b", xlab="x_i", ylab="Prob(x_i<=y)", pch=16, main="Distribution XYZ")
lines(pts, f2(pts), col="red", type="b", pch=16)
lines(pts, f3(pts), col="green", type="b", pch=16)
legend("topleft", c("x1","x2","x3"), fill=c("blue","red","green"))

enter image description here

Upvotes: 1

Related Questions