2

I am fitting a distribution to a given data.then I have estimated parameters of the distribution.I have also plotted the empirical cdf using ecdf() command in R. Now I have to plot cdf of estimated distribution along with empirical cdf. How I can do it? Can the ecdf() command still help me?

Bernhard Barker
  • 50,899
  • 13
  • 85
  • 122
user2881894
  • 21
  • 1
  • 2
  • Thanx dear, let me ask my question more specifically I have estimated parameters of "loglogistic distribution" I want to compare cdf of loglogistic distribution with emperical cdf how can i compare it visually.I have plottd the empirical cdf using ecdf().the plotting of LL cdf remains. – user2881894 Oct 15 '13 at 10:26
  • 1
    This is a coding site. Post code. – IRTFM Oct 15 '13 at 15:40

1 Answers1

1

Yes, ecdfcan help you. It has a plotting method. Below you see possible code for a normal distribution. 

x <- rnorm(100)
plot(ecdf(x))
lines(seq(-3, 3, by=.1), pnorm(seq(-3, 3, by=.1)), col=2)

EDIT: You can do the same thing using the log-logistic distribution function instead. It is implemented for instance in package actuar. 

# load package
require(actuar)
# check out the parametrization! 
?dllogis
# estimate shape and scale from your data
shape <- 20
scale <- 1
# Don't do this. Use your own data instead. 
x <- rllogis(100, shape=shape, scale=scale)
# Plotting the empirical distribution function
plot(ecdf(x))
# x-values for plotting distribution function
xvals <- seq(min(x), max(x), length=100)
# plot estimated distribution function with your estimated shape and scale
lines(xvals, pllogis(xvals, shape=shape, scale=scale), col=2)
shadow
  • 20,147
  • 4
  • 49
  • 71
  • Thanx dear, let me ask my question more specifically I have estimated parameters of "loglogistic distribution" I want to compare cdf of loglogistic distribution with emperical cdf how can i compare it visually.I have plottd the empirical cdf using ecdf().the plotting of LL cdf remains. – user2881894 Oct 15 '13 at 10:31