Data Types & Missing Data

Every Series has a single data type for all its values, and Pandas figures it out for you. Understanding that — and how a single missing value can change it — saves a lot of confusion later.

1The dtype of a Series

The dtype simply tells you what kind of data is inside. Pandas sets it automatically by looking at your values:

🔢 Whole numbers → int64
💧 Decimal numbers → float64
🔤 Text (or a mix of types) → object

dtype detective

Pandas reads your data and picks the type automatically. Watch what a single missing value does.

010

125

230

345

Pandas infers

dtype: int64

All whole numbers → Pandas uses int64.

Note

A Series is homogeneous— one dtype for the whole column. That's different from a DataFrame, where each column can have its own dtype.

2NaN — how Pandas marks a missing number

NaN ("Not a Number") comes from the NumPy library and is the main way Pandas marks a missing value. Here's the twist you saw in the widget: put a NaN into a Series of whole numbers, and the entire Series becomes float64.

nan_cast.py

import pandas as pd
import numpy as np   # np.nan lives in NumPy

s = pd.Series([10, np.nan, 30])
print(s)
# Notice: every value now ends in .0 — the dtype is float64!

import pandas as pd
import numpy as np   # np.nan lives in NumPy

s = pd.Series([10, np.nan, 30])
print(s)
# Notice: every value now ends in .0 — the dtype is float64!

Watch Out

Any maths involving NaN also results in NaN — it "infects" the calculation. NaN is Pandas' way of saying "this value is unknown", and unknown + anything is still unknown.

3Using None instead of NaN

None is Python's general-purpose "nothing" value. When you drop None into a Series of numbers, Pandas helpfully converts it to NaN for you:

none_becomes_nan.py

import pandas as pd

s2 = pd.Series([10, None, 30])
print(s2)
# None was converted to NaN, and the dtype is float64

import pandas as pd

s2 = pd.Series([10, None, 30])
print(s2)
# None was converted to NaN, and the dtype is float64

Key Takeaway

Missing numbers show up as NaN. Whether you write np.nan or None, a numeric Series stores it as NaN and quietly upgrades itself to float64.

Quick Check

You create pd.Series([1, 2, np.nan, 4]). What is its dtype?

Quick Check

What happens to None inside a numeric Series?

← Previous

🛠️ Creating a Series

Vectorized Maths ➕