w3resource

NumPy: Replace all the nan of a given array with the mean of another array

Replace NaN values with the mean of another array.

Write a NumPy program to replace all the nan (missing values) of a given array with the mean of another array.

Sample Solution:

Python Code:

# Importing the NumPy library
import numpy as np

# Creating NumPy arrays: array_nums1 from 0 to 19 reshaped into a 4x5 array and array_nums2 with NaN values
array_nums1 = np.arange(20).reshape(4, 5)
array_nums2 = np.array([[1, 2, np.nan], [4, 5, 6], [np.nan, 7, np.nan]])

# Printing the original arrays
print("Original arrays:")
print(array_nums1)
print(array_nums2)

# Replacing all the NaN values in array_nums2 with the mean of non-NaN values in array_nums1
array_nums2[np.isnan(array_nums2)] = np.nanmean(array_nums1)
print("\nAll the NaN of array_nums2 replaced by the mean of array_nums1:")
print(array_nums2)

Sample Output:

Original arrays:
[[ 0  1  2  3  4]
 [ 5  6  7  8  9]
 [10 11 12 13 14]
 [15 16 17 18 19]]
[[ 1.  2. nan]
 [ 4.  5.  6.]
 [nan  7. nan]]

All the nan of array_nums2 replaced by the mean of array_nums1:
[[1.  2.  9.5]
 [4.  5.  6. ]
 [9.5 7.  9.5]]

Explanation:

In the above example -

array_nums1 = np.arange(20).reshape(4,5) creates a 1-dimensional NumPy array containing numbers from 0 to 19 and then reshapes it into a 2-dimensional array with 4 rows and 5 columns.

array_nums2 = np.array([[1,2,np.nan],[4,5,6],[np.nan, 7, np.nan]]) creates a 2-dimensional NumPy array with NaN values.

array_nums2[np.isnan(array_nums2)]= np.nanmean(array_nums1)

In the above code -

  • np.nanmean(array_nums1): This part computes the mean of the ‘array_nums1’ while ignoring any NaN values that might be present. In this case, since there are no NaN values in array_nums1, it is equivalent to computing the mean of all elements in array_nums1.
  • array_nums2[np.isnan(array_nums2)]: This part selects all NaN values in array_nums2.
  • array_nums2[np.isnan(array_nums2)] = np.nanmean(array_nums1) replaces the selected NaN values in array_nums2 with the computed mean from array_nums1.

Python-Numpy Code Editor: