R help - Jan 2005 - Getting empirical percentiles for data

Dear List,

I have some discrete data and want to calculate the percentiles and the 
percentile ranks for each of the unique scores. I can calculate the 
percentiles with quantile().

I know that "ecdf" can be used to calculate the empirical cumulative 
distribution. However, I don't know how to exact the cumulative 
probabilities for each unique element. The requirement is similar to the 
"FREQUENCIES" in SPSS. Could someone help me in exacting the
cumulative
probabilities from the ecdf object? Thanks in advance!

# Generating artificial data
x <- round( rnorm(50, mean=50, sd=10) )
probs <- seq(0.1, 0.9, by=0.1)
# Calculating percentiles
x.percentiles <- quantile(x, probs)
# Calculating percentile ranks
x.ranks <- ecdf(x)

Best,
Mike

Dear Mike

Mike Cheung writes:
 > Dear List,
 > 
 > I have some discrete data and want to calculate the percentiles and the 
 > percentile ranks for each of the unique scores. I can calculate the 
 > percentiles with quantile().
 > 
 > I know that "ecdf" can be used to calculate the empirical
cumulative
 > distribution. However, I don't know how to exact the cumulative 
 > probabilities for each unique element. The requirement is similar to the 
 > "FREQUENCIES" in SPSS. Could someone help me in exacting the
cumulative
 > probabilities from the ecdf object? Thanks in advance!

You can use the following function:

f.freq <- function(x)
{
  tab <- data.frame(table(x))
  tab$Percent <- tab$Freq*100/length(x)
  tab$Cum.Percent[1] <- tab$Percent[1]
  for(i in 2:length(tab[,1]))
    tab$Cum.Percent[i] <- tab$Cum.Percent[i-1] + tab$Percent[i]
  tab
}

x <- round( rnorm(50, mean=50, sd=10) )
f.freq(x)

This should give you a table analog to frequencies in SPSS.


 > 
 > # Generating artificial data
 > x <- round( rnorm(50, mean=50, sd=10) )
 > probs <- seq(0.1, 0.9, by=0.1)
 > # Calculating percentiles
 > x.percentiles <- quantile(x, probs)
 > # Calculating percentile ranks
 > x.ranks <- ecdf(x)
 > 
 > Best,
 > Mike
 > 
 > ______________________________________________
 > R-help at stat.math.ethz.ch mailing list
 > https://stat.ethz.ch/mailman/listinfo/r-help
 > PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html


Best,
Christoph

-- 
Christoph Buser <buser at stat.math.ethz.ch>
Seminar fuer Statistik, LEO C11
ETH (Federal Inst. Technology)	8092 Zurich	 SWITZERLAND
phone: x-41-1-632-5414		fax: 632-1228
http://stat.ethz.ch/~buser/

Mike Cheung <mikewlcheung at hku.hk> writes:
> I know that "ecdf" can be used to calculate the empirical
cumulative
> distribution. However, I don't know how to exact the cumulative
> probabilities for each unique element. The requirement is similar to
> the "FREQUENCIES" in SPSS. Could someone help me in exacting the
> cumulative probabilities from the ecdf object? Thanks in advance!
> 
> # Generating artificial data
> x <- round( rnorm(50, mean=50, sd=10) )
> probs <- seq(0.1, 0.9, by=0.1)
> # Calculating percentiles
> x.percentiles <- quantile(x, probs)
> # Calculating percentile ranks
> x.ranks <- ecdf(x)
I don't quite see why you call it "x.ranks", but it's just a
function
that you could evaluate at the step points, so continuing your code:

val <- sort(unique(x))
cbind(val,cum.prop=x.ranks(val))

I'd go for a more direct approach though:

cumsum(prop.table(table(x)))


-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907