How to check for NaN values in Python?

To check for NaN (Not a Number) values in Python, use the following methods depending on your use case and libraries:

1. Using math.isnan() (Vanilla Python)

Checks if a float is NaN.
Works only for float-type NaN (raises TypeError for non-float inputs).

import math

x = float('nan')
print(math.isnan(x))  # True

# Handle non-float inputs safely:
y = "not a number"
try:
    print(math.isnan(y))  # Raises TypeError
except TypeError:
    print(f"{y} is not a float")

2. Using numpy.isnan() (NumPy)

Checks for NaN in NumPy arrays or scalars.
Fails for non-float/array inputs.

import numpy as np

# For scalars
x = np.nan
print(np.isnan(x))  # True

# For arrays
arr = np.array([1.0, np.nan, 3.0])
print(np.isnan(arr))  # [False  True False]

3. Using pandas.isna() (Pandas)

Best for DataFrames/Series and handles all types (NaN, None, NaT).

import pandas as pd

# For scalars
print(pd.isna(np.nan))    # True
print(pd.isna(None))      # True
print(pd.isna('text'))    # False

# For DataFrames/Series
df = pd.DataFrame({'A': [1, np.nan, None]})
print(df['A'].isna())
# 0    False
# 1     True
# 2     True

4. Check via Self-Comparison (Pure Python)

NaN is not equal to itself, so use:

x = float('nan')
print(x != x)  # True (only for NaN)

y = 5
print(y != y)  # False

---

5. Check NaN in Lists/Collections

Combine with list comprehensions or map():

import math

data = [1.0, float('nan'), 'text', None]

# Method 1: Using math.isnan (only floats)
nan_floats = [item for item in data if isinstance(item, float) and math.isnan(item)]
print(nan_floats)  # [nan]

# Method 2: Using pandas.isna (handles all types)
import pandas as pd
nan_values = [item for item in data if pd.isna(item)]
print(nan_values)  # [nan, None]

Key Differences

Method	Scope	Handles Non-Floats?	Works with Libraries
math.isnan()	Pure Python	❌ (floats only)	–
numpy.isnan()	NumPy	❌ (arrays/scalars)	NumPy
pandas.isna()	Pandas	✔️ (None, NaT, etc.)	Pandas
Self-comparison	Pure Python	✔️ (floats only)	–

---

When to Use Which

• Pandas/DataFrames: Use pd.isna().
• NumPy Arrays: Use np.isnan().
• Pure Python (Float NaN): Use math.isnan() or x != x.
• Mixed Data Types: Use pd.isna() with a try-except block.

Example: Clean a List of Mixed Data

import pandas as pd

data = [1, None, np.nan, 'text', float('nan')]
cleaned = [item for item in data if not pd.isna(item) or isinstance(item, str)]
print(cleaned)  # [1, 'text']