Typed Kriging #9
Conversation
…ill need to update tests
| struct Kriging <: AbstractBasisFunction | ||
| x | ||
| y | ||
| p |
There was a problem hiding this comment.
Do we need x y and p here? It looks to me this is being passed by the user?
One case I could think of is maybe for a plot recipe.
There was a problem hiding this comment.
I see what you are doing. I think this struct should be split into 2, Kriging and a KrigingResult and then store the results in KrigingResult.
There was a problem hiding this comment.
Thanks for your input! The idea was to store everything in Kriging because later I might get into adaptivity (I get 3 more data points, how do I change the coefficients?) and I might need the "old" x's. Not sure If I explained myself correctly.
Do you think I should create a struct Kriging() that stores x,y,p,phi and q and then pass it to the constructor KrigingResult() which has the corresponding mu,b,...,result and std_error?
There was a problem hiding this comment.
I see, your approach seems correct then 👍
There was a problem hiding this comment.
But you don't need the result pieces, right? You do k = Kriging(x::Array,y::Array,p::Number) to make it, and then do a call-overloading to do k(new_point) which returns a KrigingResult and adds the new points to the estimation?
There was a problem hiding this comment.
Not sure if I am following what you are saying, but Kriging has x,y,p,mu,b.sigma,inverse_of_R,new_point,prediction and std_error as elements of the struct. The result pieces are needed for the construction of the type Kriging. So I am not sure how I could build the struct "Kriging" without also computing those elements?
|
Merging autonomously as discussed on Slack. I will do another PR for the tests of both Kriging and Radials to keep things tidy. |
Used constructor for type Kriging, refactored code. Not tests at the moment