Nearest neighbours method

$$\hat{Z}(s_0) = \sum_{i = 1}^{k}(Z(s_i) \cdot w_i)$$

$$w_i = d_i^{0} / \sum_{i = 1}^{k}d_i^{0} = 1 / k$$

nn <- gstat(
  formula = vble ~ 1, 
  data = d2, 
  locations = ~ x + y, 
  nmax = 5, 
  set = list(idp = 0)
)