Efficient way to write a python dictionary with Numpy Nd-Array values into a Json File

Question

How do I efficiently write a Python dictionary, where the values are Numpy Nd-Arrays to a Json File? I obtain an error saying that the Numpy Nd-Array is not Json-Serializable. Is there any way to overcome this?

Answer 1

JSON only supports a limited number of datatypes. If you want to store other types of data as JSON then you need to convert it to something that JSON accepts. The obvious choice for Numpy arrays is to store them as (possibly nested) lists. Fortunately, Numpy arrays have a .tolist method which performs the conversion efficiently.

import numpy as np
import json

a = np.array(range(25), dtype=np.uint8).reshape(5, 5) 
print(a)
print(json.dumps(a.tolist()))

output

[[ 0  1  2  3  4]
 [ 5  6  7  8  9]
 [10 11 12 13 14]
 [15 16 17 18 19]
 [20 21 22 23 24]]
[[0, 1, 2, 3, 4], [5, 6, 7, 8, 9], [10, 11, 12, 13, 14], [15, 16, 17, 18, 19], [20, 21, 22, 23, 24]]

.tolist will convert the array elements to native Python types (int or float) if it can do so losslessly. If you use other datatypes I suggest you convert them to something portable before calling .tolist.