CODEWITHSUNDEEP
rpdf
Lio
Lio

Reputation: 111

Comparing and vizualizing densities of two samples in R

I have to compare the density function of two samples in R. Surprisingly, whatever the function I use, plot(), lines() or ggplot, one of both samples either does not appear or both areas are different and cannot be equal to one. I would like both areas equal one on the same graph as to determine easily the set of abscissa values for which the pdf of a sample is larger than the pdf of the other. How can I solve it? Many thanks for your help.

1/ Using ggplot, the script is: 

require ("ggplot2")
p2<-density(tabgroupcl2$B, n=1000)
p1<-density(tabgroupcl1$B, n=1000)
dat <- data.frame(dens = c(p1$x, p2$x)
                   , lines = rep(c("cl1", "cl2")), each=1000)
ggplot(dat,aes(x = dens, fill = lines)) + geom_density(alpha = 0.5)

2/ Where Density(tabgroupcl2$B):

Call:
        density.default(x = tabgroupcl2$B)

Data: tabgroupcl2$B (348 obs.); Bandwidth 'bw' = 0.001689

       x                y           
 Min.   :-91.95   Min.   :  0.0000  
 1st Qu.:-34.07   1st Qu.:  0.0000  
 Median : 23.80   Median :  0.0000  
 Mean   : 23.80   Mean   :  0.4613  
 3rd Qu.: 81.68   3rd Qu.:  0.0000  
 Max.   :139.56   Max.   :179.2431

3/ Where Density(tabgroupcl1$B):

Call:
        density.default(x = tabgroupcl1$B)

Data: tabgroupcl1$B (9 obs.);   Bandwidth 'bw' = 0.2738

       x                y            
 Min.   :-2.607   Min.   :0.0000000  
 1st Qu.: 1.495   1st Qu.:0.0000000  
 Median : 5.598   Median :0.0001349  
 Mean   : 5.598   Mean   :0.0608673  
 3rd Qu.: 9.700   3rd Qu.:0.0548682  
 Max.   :13.802   Max.   :0.7583033

Upvotes: 0

Views: 1574

Answers (3)

KLDavenport
KLDavenport

Reputation: 679

enter image description hereThis is how I overlayed two datasets columns in one plot:

Here I am creating a new column (samediff) in each dataframe with repeating text in the column to identify it. same_auditor$samediff <- "same" diff_auditor$samediff <- "diff"

Combining the datasets for the plot. samescore <- same_auditor$Audit_Score diffscore <- diff_auditor$Audit_Score sameDiffcombined <- rbind(diff_auditor,same_auditor)

ggplot(sameDiffcombined, aes(Audit_Score, fill = samediff)) + geom_density(alpha = 0.5)

Upvotes: 0

AdamO
AdamO

Reputation: 4930

Peter Ellis' answer is exactly right. Here's a "ground up" example of how you might estimate and plot densities from two different samples on the same axis:

x <- rnorm(1000, mean=3, sd=2)
y <- rnorm(500, mean=3.5, sd=3)

dx <- density(x)
dy <- density(y)

plot.new()
plot.window(xlim=range(c(dx$x, dy$x)), ylim=range(c(dx$y, dy$y)))
with(dx, lines(x, y))
with(dy, lines(x, y, lty=2))
axis(1)
axis(2)
legend(topright, lty=1:2, c('x', 'y'))
mtext(side=1, line=2, 'Observed values')
mtext(side=2, line=2, 'Estimated probability mass')
title('Smoothed Density Estimates for 2-sample experiment')

Upvotes: 1

Peter Ellis
Peter Ellis

Reputation: 5904

It looks like you are using the x values from a density object in your ggplot() function as though they were the original data. I don't see why you would want to do this, but if so you also need to use the y values - and you don't need the density stat in ggplot at all. Alternatively, let the density stat do the work and use your original data.

Upvotes: 1

Related Questions