如何在对数刻度上绘制CCDF图? [英] How to plot CCDF graph on a logarithmic scale?

问题描述

我想在对数对数轴上绘制一些模拟的幂律尾部数据的CCDF图，下面是在法线轴上绘制CCDF图的R代码，我在链接上使用了代码: (如何绘制CCDF图形?)

I want to plot a CCDF graph for some of my simulated power-law tail data on a log-log axis, below is my R code of plotting a CCDF graph on a normal axis, I used the code on the link: (How to plot a CCDF gragh?)

> load("fakedata500.Rda")
> x<-fakedata500
> f<-ecdf(x)
> f
Empirical CDF 
Call: ecdf(x)
 x[1:500] = 0.50174, 0.50307, 0.50383,  ..., 81.674, 140.63
> plot(f)

下面是ECDF图:

> plot(sort(x), 1-f(sort(x)), type="s", lwd=1)

此命令为我提供了CCDF图:

and this command gives me the CCDF graph:

但是，我想在对数-对数轴上绘制CCDF图，以生成如下图所示的图: (摘自最小化识别Lévy飞行行为的错误".)

However, I would like to plot my CCDF graph on a log-log axis to produce a graph like the picture below: (graph from "Minimizing errors in identifying Lévy flight behaviour of organisms.")

有没有办法在R中做到这一点?

Is there a way to do it in R?

如果是这样，如何在CCDF图上进行线性回归?我已经尝试过使用下面的命令，但这对我不起作用.

If so, how to do a linear regression on the CCDF graph as well? I have tried using command below but that just not working for me.

a<-plot(sort(x), 1-f(sort(x)), type="s", lwd=1)
> a
NULL
> res=lm(a)
Error in terms.formula(formula, data = data) : 
  argument is not a valid model

非常感谢.

更新:

我使用了@BondedDust提供的代码，并成功生成了CCDF图:

I used the code given by @BondedDust and successfully generated a CCDF graph:

(plot(sort(x) , 1-ecdf(x)(sort(x) ), log="xy"))

下面是我如何生成数据集的代码:

Below is the code of how I generated my dataset:

u<-runif(500)
fakedata500<-((2*(1-u))^(-1))

推荐答案

这是另一种使用分位数的方法.

Here's another way using quantiles.

library(VGAM)  # for rpareto(...)
set.seed(1)    # for reproducible example
X <- rpareto(1000,location=1,shape=1)
p <- ppoints(100)
par(mfrow=c(1,3))
plot(quantile(X,p=p),p,type="l",ylab="P(X < x)",xlab="x",main="CDF")
plot(quantile(X,p=p),1-p,type="l",ylab="P(X > x)",xlab="x",main="CCDF")
plot(log(quantile(X,p=p)),log(1-p),
     ylab="log[P(X > x)]",xlab="log(x)",main="CCDF: log-log")

这是回归.

df  <- data.frame(x=log(1-p),y=log(quantile(X,p=p)))
fit <- lm(y~x,df)
summary(fit)
# ...
# Coefficients:
#              Estimate Std. Error  t value Pr(>|t|)    
# (Intercept)  0.039559   0.007584    5.216 1.02e-06 ***
# x           -0.944380   0.005427 -174.028  < 2e-16 ***
# ---
# Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
# 
# Residual standard error: 0.05317 on 98 degrees of freedom
# Multiple R-squared:  0.9968,  Adjusted R-squared:  0.9967 
# F-statistic: 3.029e+04 on 1 and 98 DF,  p-value: < 2.2e-16

这篇关于如何在对数刻度上绘制CCDF图?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！