Fill in missing values with nearest neighbour in Python numpy masked arrays?

user_413363 0
Viewed 12815

I am working with a 2D Numpy masked_array in Python. I need to change the data values in the masked area such that they equal the nearest unmasked value.

NB. If there are more than one nearest unmasked values then it can take any of those nearest values (which ever one turns out to be easiest to code…)

e.g.

import numpy
import numpy.ma as ma

a = numpy.arange(100).reshape(10,10)
fill_value=-99
a[2:4,3:8] = fill_value
a[8,8] = fill_value
a = ma.masked_array(a,a==fill_value)

>>> a  [[0 1 2 3 4 5 6 7 8 9]
  [10 11 12 13 14 15 16 17 18 19]
  [20 21 22 -- -- -- -- -- 28 29]
  [30 31 32 -- -- -- -- -- 38 39]
  [40 41 42 43 44 45 46 47 48 49]
  [50 51 52 53 54 55 56 57 58 59]
  [60 61 62 63 64 65 66 67 68 69]
  [70 71 72 73 74 75 76 77 78 79]
  [80 81 82 83 84 85 86 87 -- 89]
  [90 91 92 93 94 95 96 97 98 99]],
  • I need it to look like this:
>>> a.data
 [[0 1 2 3 4 5 6 7 8 9]
 [10 11 12 13 14 15 16 17 18 19]
 [20 21 22 ? 14 15 16 ? 28 29]
 [30 31 32 ? 44 45 46 ? 38 39]
 [40 41 42 43 44 45 46 47 48 49]
 [50 51 52 53 54 55 56 57 58 59]
 [60 61 62 63 64 65 66 67 68 69]
 [70 71 72 73 74 75 76 77 78 79]
 [80 81 82 83 84 85 86 87 ? 89]
 [90 91 92 93 94 95 96 97 98 99]],

NB. where "?" could take any of the adjacent unmasked values.

What is the most efficient way to do this?

Thanks for your help.

scipy numpy python
3 Answers

For more complicated cases you could use scipy.spatial:

from scipy.spatial import KDTree
x,y=np.mgrid[0:a.shape[0],0:a.shape[1]]

xygood = np.array((x[~a.mask],y[~a.mask])).T
xybad = np.array((x[a.mask],y[a.mask])).T

a[a.mask] = a[~a.mask][KDTree(xygood).query(xybad)[1]]

print a
  [[0 1 2 3 4 5 6 7 8 9]
  [10 11 12 13 14 15 16 17 18 19]
  [20 21 22 13 14 15 16 17 28 29]
  [30 31 32 32 44 45 46 38 38 39]
  [40 41 42 43 44 45 46 47 48 49]
  [50 51 52 53 54 55 56 57 58 59]
  [60 61 62 63 64 65 66 67 68 69]
  [70 71 72 73 74 75 76 77 78 79]
  [80 81 82 83 84 85 86 87 78 89]
  [90 91 92 93 94 95 96 97 98 99]]
user_12691400
Related