Rex_Bryan@urscorp.com
2003-Jan-24 20:11 UTC
[R] The other type of relative frequency for hist()
Hist() can plot histogram bars with a relative frequency scale. However, this relative frequency is scaled as a probability density, where the sum of the bar heights times the bar widths will equal 1. There is another way to define relative frequency which is based on the proportion of the count in a bin to the total count over all bins. Is there a way to scale the vertical axis so as to show this other type? Example: v3<-c(0.03155,0.073,0.280,0.285,0.285,0.300,0.035,0.035,0.035,0.035,0.04025) length(v3) [1] 11 hist.dump<-hist(v3, freq=FALSE, plot=FALSE) hist.dump$count [1] 6 1 0 0 0 4 hist.dump$density [1] 10.909078 1.818182 0.000000 0.000000 0.000000 7.272727 Now for this histogram 3 bins have bars with widths in this case that happen to be 0.05 units wide. If hist.dump$count is multiplied by hist.dump$density for the first bin a value of 0.5455 is obtained. This is of course simply the 6 observations in the first bin divided by 11. Is there a simple way of representing this second type of relative frequency in the y-axis? REX