如何解决错误因素在 R 中的错误级别

Question

I have some difficulties applying the inhomogeneous G-function to my point pattern in R . In order to use GmultiInhom , I first tried to convert my point pattern bci.tree8pppa to a multitype pattern:

bci.tree8multi = ppp(bci.tree8pppa$x, bci.tree8pppa$y, window=owin(c(0,1000), c(0,500)), marks = factor(bci.tree8pppa$marks[,3]))

Then applied the G-function as follows:

G = GmultiInhom(bci.tree8multi, marks(bci.tree8multi) == species1, marks(bci.tree8multi) == species2, lambdaI = lambda1points, lambdaJ = lambda2points, lambdamin = min(lambda2points), r = c(0,r1,r2,r3))

But this yields the error: "Error in split.default(X, group): factor has bad level"

How can I solve this? Thank you in advance!

Answer 1

For the benefit of any R programmers out there: I traced the error message "factor has bad level" to the C source code for .Internal(split.default(x,f)) in the base R system. This error message can occur only when f is a list rather than a factor . The code converts f to a factor using the function interaction which performs character string manipulations. This conversion can go wrong, in the sense that the resulting factor has "bad levels": the integer representation of the factor includes values less than 1 or greater than the number of levels of the factor. Then the error occurs.

The original post has not provided a working example, so it's difficult to figure out exactly how the code in spatstat::GmultiInhom caused the wrong type of data to be fed to split.default . However, it must be related to the misuse of the argument r . The code in spatstat will be tightened to enforce stricter requirements on the format of r .