JIYIK CN >

This article discusses some built-in NumPy functions that you can use to remove nanvalues.

Remove Nan values ​​using logical_not()and methods in NumPyisnan()

logical_not() is used to apply logical NOT to the elements of an array. isnan()It is a Boolean function that checks whether an element is nan.

Using isnan()the function, we can create a Boolean array that nanhas all non- values ​​of Falseand all nanvalues ​​of True. Next, using logical_not()the function, we can Trueconvert from to Falseand vice versa.

Finally, using boolean indexing, we can filter all non- values ​​from the original NumPy array nan. All Trueindices with as their value will be used to filter the NumPy array.

For an in-depth look at these functions, please refer to their official documentation, here and here, respectively.

Refer to the following code snippet for the solution.

import numpy as np

myArray = np.array([1, 2, 3, np.nan, np.nan, 4, 5, 6, np.nan, 7, 8, 9, np.nan])
output1 = myArray[np.logical_not(np.isnan(myArray))]  # Line 1
output2 = myArray[~np.isnan(myArray)]  # Line 2
print(myArray)
print(output1)
print(output2)

Output:

[ 1.  2.  3. nan nan  4.  5.  6. nan  7.  8.  9. nan]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]

Line 2is Line 1a simplified version of .

isfinite()Remove Nan values ​​using method in NumPy

As the name suggests, isfinite()the function is a boolean function that checks if an element is a finite value. It can also check for finite values ​​in an array and return a boolean array for that array. The boolean array will nanstore for all values False, for all finite values True.

We will use this function to retrieve a boolean array for the target array. Using boolean indexing, we will filter all finite values. Again, as mentioned above, Truethe index with the value will be used to filter the array.

Here is the sample code.

import numpy as np

myArray1 = np.array([1, 2, 3, np.nan, np.nan, 4, 5, 6, np.nan, 7, 8, 9, np.nan])
myArray2 = np.array([np.nan, np.nan, np.nan, np.nan, np.nan, np.nan])
myArray3 = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
output1 = myArray1[np.isfinite(myArray1)]
output2 = myArray2[np.isfinite(myArray2)]
output3 = myArray3[np.isfinite(myArray3)]
print(myArray1)
print(myArray2)
print(myArray3)
print(output1)
print(output2)
print(output3)

Output:

[ 1.  2.  3. nan nan  4.  5.  6. nan  7.  8.  9. nan]
[nan nan nan nan nan nan]
[ 1  2  3  4  5  6  7  8  9 10]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]
[]
[ 1  2  3  4  5  6  7  8  9 10]

To learn more about this function, refer to the official documentation

Use math.isnanthe method to remove Nan values

In addition to these two NumPy solutions, there are two other nanways to remove a value. These two ways involve the function mathin the library isnan()and the function pandasin the library isnull. Both functions check whether an element is nan, and return a Boolean result.

Here is isnan()a solution using the method.

import numpy as np
import math

myArray1 = np.array([1, 2, 3, np.nan, np.nan, 4, 5, 6, np.nan, 7, 8, 9, np.nan])
myArray2 = np.array([np.nan, np.nan, np.nan, np.nan, np.nan, np.nan])
myArray3 = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
booleanArray1 = [not math.isnan(number) for number in myArray1]
booleanArray2 = [not math.isnan(number) for number in myArray2]
booleanArray3 = [not math.isnan(number) for number in myArray3]
print(myArray1)
print(myArray2)
print(myArray3)
print(myArray1[booleanArray1])
print(myArray2[booleanArray2])
print(myArray3[booleanArray3])

Output:

[ 1.  2.  3. nan nan  4.  5.  6. nan  7.  8.  9. nan]
[nan nan nan nan nan nan]
[ 1  2  3  4  5  6  7  8  9 10]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]
[]
[ 1  2  3  4  5  6  7  8  9 10]

Use pandas.isnullthe method to remove Nan values

Here is a solution using the method pandasfrom .isnull()

import numpy as np
import pandas as pd

myArray1 = np.array([1, 2, 3, np.nan, np.nan, 4, 5, 6, np.nan, 7, 8, 9, np.nan])
myArray2 = np.array([np.nan, np.nan, np.nan, np.nan, np.nan, np.nan])
myArray3 = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
booleanArray1 = [not pd.isnull(number) for number in myArray1]
booleanArray2 = [not pd.isnull(number) for number in myArray2]
booleanArray3 = [not pd.isnull(number) for number in myArray3]
print(myArray1)
print(myArray2)
print(myArray3)
print(myArray1[booleanArray1])
print(myArray2[booleanArray2])
print(myArray3[booleanArray3])
print(myArray1[~pd.isnull(myArray1)])  # Line 1
print(myArray2[~pd.isnull(myArray2)])  # Line 2
print(myArray3[~pd.isnull(myArray3)])  # Line 3

Output:

[ 1.  2.  3. nan nan  4.  5.  6. nan  7.  8.  9. nan]
[nan nan nan nan nan nan]
[ 1  2  3  4  5  6  7  8  9 10]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]
[]
[ 1  2  3  4  5  6  7  8  9 10]
[1. 2. 3. 4. 5. 6. 7. 8. 9.]
[]
[ 1  2  3  4  5  6  7  8  9 10]