Hello,
I'm trying to calculate non-parametric probabilities using the np package
and having some difficulties.
OS is Windows, R version 2.11.1
Here is what I've done so far.
library(np)
veg <- data.frame(factor(Physiogomy), meanAnnualDepthAve, TP)
attach(veg) : for clarification dim(veg) returns 1292 3
fy.x <- npcdens(veg$factor.Physiogomy ~ veg$meanAnnualDepthAve, nmulti=1)
# this works, but I haven't found any information explaining what the
nmulti=1 term is doing? Does this set the number of levels in the factor?
My data actually has 8 types, can I develop this to treat each one in a
single function ?
veg.eval <- data.frame(Physiogomy = factor('Marl"},
meanAnnualDepthAve seq(min(meanAnnualDepthAve), max(meanAnnualDepthAve))
# This also works, however where does the 4755 records originate
str(veg.eval)
' data.frame': 4755 obs of 2 variables
$ Physiogomy : factor w / 1 level "Marl" : 1
1
1 1 ....
$ meanAnnualDepthAve : num -592, -591 - 590 - 578
because the data frame veg only contains 1292 records the is a mismatch
between the 4755 records. Why are so many records produced in the veg.eval
statement and how can i constrain it to be consistent with the dimensions of
veg ?
plot(x, y, type = "l", lty="2", col='red' , xlab =
"Mean Annual Depth",
ylab="Estimated Prob of Marl")
lines(veg.eval$meanAnnualDepthAve, predict(fy.x, newdata=veg.eval),
col='blue')
I'm following an example I found Here:
http://en.wikipedia.org/wiki/Density_estimation
Your help is greatly appreciated.
Thanks
Steve
[[alternative HTML version deleted]]
