Reindexing a Series

Reindexing is a technique for creating a brand-new Series from an existing one — the new Series has the same kind of labels, but arranged in the order you specify.

1Reorder the labels

reindex() takes a list of index labels and returns a new Series whose rows follow that exact order. Explore what happens when you reorder, reverse, or ask for a label that doesn't exist:

Reindex explorer

Reindexing builds a BRAND-NEW Series with the labels you ask for, in the order you ask for. Ask for a label that doesn't exist and its value is NaN.

s2 = s1.reindex(['b', 'd', 'c', 'a', 'e'])

s1 (original)

ahi

bhello

cbye

dok

ehappy

→

s2 (reindexed)

bhello

dok

cbye

ahi

ehappy

Same values, new order — and s1 itself never changes. reindex() always returns a new Series.

2Reordering existing labels

Pass the labels in whatever order you want. The values travel with their labels, and the original Series is left untouched.

reindex.py

import pandas as pd

s1 = pd.Series(['hi', 'hello', 'bye', 'ok', 'happy'],
               index=['a', 'b', 'c', 'd', 'e'])

# A NEW Series with the same labels in a new order
s2 = s1.reindex(['b', 'd', 'c', 'a', 'e'])
print(s2)

print("\n--- s1 is unchanged ---")
print(s1)

import pandas as pd

s1 = pd.Series(['hi', 'hello', 'bye', 'ok', 'happy'],
               index=['a', 'b', 'c', 'd', 'e'])

# A NEW Series with the same labels in a new order
s2 = s1.reindex(['b', 'd', 'c', 'a', 'e'])
print(s2)

print("\n--- s1 is unchanged ---")
print(s1)

3Asking for a label that doesn't exist

If you list a label that isn't in the original Series, Pandas still creates that row — but it has no value to put there, so it fills it with NaN (which also makes the Series float64 if it was numeric).

reindex_nan.py

import pandas as pd

marks = pd.Series([10, 20, 30], index=['a', 'b', 'c'])

# 'z' doesn't exist in marks → its value becomes NaN
s = marks.reindex(['a', 'b', 'z', 'c'])
print(s)

import pandas as pd

marks = pd.Series([10, 20, 30], index=['a', 'b', 'c'])

# 'z' doesn't exist in marks → its value becomes NaN
s = marks.reindex(['a', 'b', 'z', 'c'])
print(s)

Tip

Want a different filler than NaN? Pass fill_value=0 (or any value): marks.reindex(['a', 'z'], fill_value=0).

Key Takeaway

reindex() always returns a new Series in the order you asked for. Labels that don't exist come back as NaN (or your fill_value).

Quick Check

s has labels a, b, c. What does s.reindex(['c', 'a']) return?

Quick Check

You reindex a numeric Series with a label that doesn't exist. That row's value is…

← Previous

🎯 Selection, Indexing & Slicing

Dropping Entries 🗑️