R: how to obtain specific density estimates

Question

# The Old Faithful geyser data
d <- density(faithful$eruptions, bw = "sj")
> head(d$x)
[1] 1.179869 1.188363 1.196857 1.205350 1.213844 1.222338

I'm using density function in {stats}, and I'm wondering if it's possible to see density at specific values in the output? For example, currently, I have density estimates at eruption values of [1] 1.179869 1.188363 ... but what if I want to know the density estimates at eruption values 1 2 5 10 ...? Is there a way to extract these the density object, d?

you might have to use `interp` on the `x` and `y` components of the returned object ... — Ben Bolker, Aug 03 '17 at 03:20
are you looking for something like this: `x <- 2; approx(d$x,d$y,xout = x)`? — AK88, Aug 03 '17 at 03:25
If you mean that you want to extract the x values at a given location in the data set, then you can just use `d$x[3]` for a the third value. or `d$y[5]` for the fifth location y result, you can extract any value you want this way. — sconfluentus, Aug 03 '17 at 03:38
@AK88, yeah, something like that. Do you know how off the linear interpolation would be from the actual density estimate provided by the function? — Adrian, Aug 03 '17 at 04:29
I think `approx` will be fine. Even the documentation of `stats::density` mentions that they will use `approx` if you request to compute the density at more then 512 points. — AEF, Aug 03 '17 at 08:16

DataTx · Accepted Answer · 2018-01-15T14:50:37.853

If I understand you correctly, you want the probabilities where the x value equals some number (3 or 4 as in my solution)?

d <- density(faithful$eruptions, bw = "sj")
densityDF <- data.frame(xVals = d$x, prob = d$y)
densityDF$xVals <- round(densityDF$xVals)

densitySearch <- densityDF[densityDF$xVals %in% c(3,4),]

Result:

   xVals       prob
157     3 0.11229482
158     3 0.10721410
159     3 0.10230912
160     3 0.09765156
161     3 0.09318662
162     3 0.08891621

R: how to obtain specific density estimates

1 Answers1