Filling Missing Data

Why Fill Instead of Drop?

Dropping removes entire rows. Filling keeps your data and replaces empty cells with something useful.

Fill with a Fixed Value

code.py

import pandas as pd
import numpy as np

df = pd.DataFrame({
    'Name': ['John', 'Sarah', None],
    'Age': [25, None, 30],
    'Score': [85, 90, None]
})

# Fill all missing with 0
filled = df.fillna(0)
print(filled)

Output:

   Name   Age  Score
0  John  25.0   85.0
1 Sarah   0.0   90.0
2     0  30.0    0.0

Fill Different Values for Different Columns

code.py

filled = df.fillna({
    'Name': 'Unknown',
    'Age': 0,
    'Score': 50
})
print(filled)

Output:

      Name   Age  Score
0     John  25.0   85.0
1    Sarah   0.0   90.0
2  Unknown  30.0   50.0

Fill with Average (Mean)

Best for numbers. Keeps the overall average same.

code.py

df = pd.DataFrame({
    'Score': [80, 90, None, 85, None]
})

# Fill with average
avg = df['Score'].mean()
df['Score'] = df['Score'].fillna(avg)
print(df)

Output:

   Score
0   80.0
1   90.0
2   85.0  <- was missing, now average
3   85.0
4   85.0  <- was missing, now average

Fill with Middle Value (Median)

Better when data has extreme values.

code.py

df['Score'] = df['Score'].fillna(df['Score'].median())

Fill with Most Common Value (Mode)

Best for categories like "Male/Female" or "Yes/No".

code.py

df = pd.DataFrame({
    'City': ['NYC', 'LA', None, 'NYC', 'NYC']
})

# Most common city
most_common = df['City'].mode()[0]
df['City'] = df['City'].fillna(most_common)
print(df)

Output:

  City
0  NYC
1   LA
2  NYC  <- was missing, now most common
3  NYC
4  NYC

Fill with Previous Value (Forward Fill)

Good for time data. Uses the value before the empty cell.

code.py

df = pd.DataFrame({
    'Day': [1, 2, 3, 4, 5],
    'Temp': [20, None, None, 25, 26]
})

df['Temp'] = df['Temp'].ffill()
print(df)

Output:

   Day  Temp
0    1  20.0
1    2  20.0  <- copied from row 0
2    3  20.0  <- copied from row 1
3    4  25.0
4    5  26.0

Fill with Next Value (Backward Fill)

Uses the value after the empty cell.

code.py

df['Temp'] = df['Temp'].bfill()

Which Method to Use?

Data Type	Best Method
Numbers (normal)	Mean
Numbers (has outliers)	Median
Categories	Mode
Time series	Forward/Backward fill
Unknown	Fixed value like 0 or "Unknown"

Key Points

fillna(value) replaces all missing with one value
fillna({'col': value}) different values per column
fillna(df['col'].mean()) fills with average
ffill() fills with previous value
bfill() fills with next value

What's Next?

For time data, there's a smarter way to fill: interpolation. Learn it next!