Replace NaN with zero and fill negative infinity values in Python

In NumPy, you can handle NaN and infinity values using the numpy.nan_to_num() method. This function replaces NaN with zero and infinity values with large finite numbers, making your data suitable for mathematical operations.

Syntax

numpy.nan_to_num(x, copy=True, nan=0.0, posinf=None, neginf=None)

Parameters

Basic Example

Let's create an array with NaN and infinity values and clean them ?

import numpy as np

# Create array with problematic values
arr = np.array([np.inf, -np.inf, np.nan, -128, 128])
print("Original Array:")
print(arr)

# Replace NaN and infinity values
cleaned = np.nan_to_num(arr)
print("\nCleaned Array:")
print(cleaned)

Original Array:
[ inf -inf  nan -128.  128.]

Cleaned Array:
[ 1.79769313e+308 -1.79769313e+308  0.00000000e+000 -1.28000000e+002
  1.28000000e+002]

Custom Replacement Values

You can specify custom values for NaN and infinity replacements ?

import numpy as np

arr = np.array([np.inf, -np.inf, np.nan, 5.5, -3.2])
print("Original Array:")
print(arr)

# Custom replacement values
result = np.nan_to_num(arr, nan=999, posinf=1000, neginf=-1000)
print("\nWith Custom Values:")
print(result)

Original Array:
[ inf -inf  nan  5.5 -3.2]

With Custom Values:
[ 1.00000000e+03 -1.00000000e+03  9.99000000e+02  5.50000000e+00
 -3.20000000e+00]

In-place Modification

Use copy=False to modify the original array directly ?

import numpy as np

arr = np.array([np.nan, np.inf, -np.inf, 42.0])
print("Before modification:")
print(arr)
print("Array ID:", id(arr))

# Modify in-place
np.nan_to_num(arr, copy=False, nan=0, posinf=100, neginf=-100)
print("\nAfter in-place modification:")
print(arr)
print("Array ID:", id(arr))  # Same ID confirms in-place operation

Before modification:
[ nan  inf -inf  42.]
Array ID: 140234567890123

After in-place modification:
[   0.  100. -100.   42.]
Array ID: 140234567890123

Practical Use Case

This is especially useful when cleaning data for machine learning ?

import numpy as np

# Simulated sensor data with missing/invalid readings
sensor_data = np.array([23.5, np.nan, np.inf, 24.1, -np.inf, 22.8, np.nan])
print("Raw sensor data:")
print(sensor_data)

# Clean the data for analysis
clean_data = np.nan_to_num(sensor_data, nan=23.0, posinf=30.0, neginf=15.0)
print("\nCleaned sensor data:")
print(clean_data)
print(f"Mean temperature: {clean_data.mean():.2f}")

Raw sensor data:
[23.5  nan  inf 24.1 -inf 22.8  nan]

Cleaned sensor data:
[23.5 23.  30.  24.1 15.  22.8 23. ]
Mean temperature: 23.06

Conclusion

The numpy.nan_to_num() function is essential for data preprocessing, converting problematic NaN and infinity values into workable numbers. Use custom replacement values when you need specific behavior for your data analysis.