Nearest Neighbor Distances for a 2D Grid in Python

Question

I have a data set of discrete points over a 2D grid with dimensions of 4x4, each have a given x and y coordinate such as:

x_coord, y_coord
0.5     2.1
0.8     3.2
3.3     1.1
2.8     0.6
1.9     1.0

I want to step through each grid point to find the distance to the third nearest neighbor. So I want to know the distance to the third nearest neighbor for the following coordinates:

[0,0]
[0,1]
[0,2]
[0,3]
[1,0]
[1,1]
[1,2]
[1,3]
[2,0]
[2,1]
[2,2]
[2,3]
[3,0]
[3,1]
[3,2]
[3,3]

I think that my trouble is coming from how to step through a grid and do the calculation for the distance to the nearest neighbor. I am just learning python so my experience with nesting for loops is very limited, but my guess is that's what I need here.

So far I have

from sklearn.neighbors import NearestNeighbors
import numpy as np

x=[0.5,0.8,3.3,2.8,1.9]
y=[2.1,3.2,1.1,0.6,1.0]

myarray = np.array([x,y])
myarray_trans = np.transpose(myarray)
myarray_trans

nbrs = NearestNeighbors(n_neighbors=3, algorithm='ball_tree').fit(myarray_trans)
distances, indices = nbrs.kneighbors(myarray_trans)

Now using this as my training set I need to step through the full grid to find the distance to the third nearest neighbor from the discrete list. I have tried this:

for i in range(0,4):
   for j in range(0,4):
      distances2[i,j], indices[i,j] = nbrs.kneighbors([i,j])

I get an error that says "Expected 2D array, got 1D array instead: array=[0 0]

I will have to do this on a bigger data set with a grid of something more like the size of 500 x 500 so this is an example.

Hey, you better give use the stack when you bump on an error, so we can see which call fail rather than guessing from what you say. — ninjaconcombre

ninjaconcombre ninjaconcombre · Accepted Answer · 2021-04-10T08:51:43

From what I can see on https://scikit-learn.org/stable/modules/neighbors.html, nbrs.kneighbors expect a list of list as input. This is precisely what the error is telling you.

In python nbrs.kneighbors([i,j]) evaluate to a list such as [3,8], which is a simple list and not a multidimensional array

Nearest Neighbor Distances for a 2D Grid in Python

1 Answers