apply, map & applymap

Reviewed & published by Brayan K

apply, map, and applymap are pandas methods that run your own function across data: map transforms a Series element by element, apply runs a function over a Series or over the rows and columns of a DataFrame, and applymap (now DataFrame.map) touches every single cell.

Part of the free Pandas course at LearnCodingFast — hands-on lessons with examples you run in your browser, plus practice exercises and a quick quiz.

Learn which tool to reach for, how the axis argument changes everything, and how lambdas keep your transformations short.

What You'll Learn in This Lesson

• Transform a Series with Series.apply(func)
• Recode values with Series.map({dict})
• Run a function down columns with axis=0
• Run a function across rows with axis=1
• Touch every cell with applymap / DataFrame.map
• Write compact lambda functions

1 Series.apply() and Series.map()

On a single column, Series.apply(func) calls your function once per value and collects the results into a new Series. Series.map does the same when given a function, but it can also take a dictionary to recode values, which is perfect for relabelling categories.

import pandas as pd

prices = pd.Series([10.0, 25.5, 8.99])

# apply a function to every value
with_tax = prices.apply(lambda p: round(p * 1.2, 2))
print(with_tax)
# 12.0, 30.6, 10.79

codes = pd.Series(["S", "M", "L", "S"])

# map with a dictionary to recode each value
labels = codes.map({"S": "Small", "M": "Medium", "L": "Large"})
print(labels.tolist())
# ['Small', 'Medium', 'Large', 'Small']

# map can also take a function, just like apply
print(codes.map(lambda c: c.lower()).tolist())
# ['s', 'm', 'l', 's']

# ✅ Expected output:
# 0    12.00
# 1    30.60
# 2    10.79
# dtype: float64
# ['Small', 'Medium', 'Large', 'Small']
# ['s', 'm', 'l', 's']

2 DataFrame.apply: axis=0 vs axis=1

On a DataFrame, apply hands your function a whole Series at a time. With axis=0 (the default) each column is passed in, so you get one summary per column. With axis=1 each row is passed in, which is how you combine several columns into one new value.

import pandas as pd

df = pd.DataFrame({
    "math": [80, 90, 70],
    "science": [60, 85, 95],
})

# axis=0: function receives each COLUMN -> one result per column
print(df.apply(lambda col: col.max() - col.min()))
# math -> 20, science -> 35

# axis=1: function receives each ROW -> one result per row
df["total"] = df.apply(lambda row: row["math"] + row["science"], axis=1)
print(df)
# total column: 140, 175, 165

# ✅ Expected output:
# math       20
# science    35
# dtype: int64
#    math  science  total
# 0    80       60    140
# 1    90       85    175
# 2    70       95    165

3 Elementwise: applymap (now DataFrame.map)

When you want to transform every individual cell of a DataFrame the same way, use applymap. In pandas 2.1 and later this method was renamed DataFrame.map; both do exactly the same elementwise work, so you may see either name.

import pandas as pd

df = pd.DataFrame({
    "a": [1.5, 2.25],
    "b": [3.1, 4.999],
})

# Format every cell as a 2-decimal string.
# applymap works on all pandas versions...
print(df.applymap(lambda x: f"{x:.2f}"))

# ...and DataFrame.map is the modern equivalent (pandas 2.1+)
print(df.map(lambda x: f"{x:.2f}"))
# Both print:
#       a     b
# 0  1.50  3.10
# 1  2.25  5.00

# ✅ Expected output:
#       a     b
# 0  1.50  3.10
# 1  2.25  5.00
#       a     b
# 0  1.50  3.10
# 1  2.25  5.00

🎯 Your Turn: Pass or Fail

Replace the blank so each score becomes "Pass" when it is 50 or more and "Fail" otherwise. The expected output is a Series with values Pass, Fail, Pass.

import pandas as pd

scores = pd.Series([72, 41, 88])

# 🎯 YOUR TURN: return "Pass" if s >= 50 else "Fail"
result = scores.apply(lambda s: "Pass" if ___ else "Fail")
print(result.tolist())

# Expected output: ['Pass', 'Fail', 'Pass']

Common Errors

❌ Forgetting axis=1 for row logic

Without axis=1, apply walks columns and your row lookup like row["math"] fails or gives the wrong shape.

df.apply(lambda row: row["a"] + row["b"], axis=1)  # ✅ axis=1

❌ Using applymap on a Series

applymap and DataFrame.map only exist on DataFrames. On a single column use Series.apply or Series.map instead.

series.apply(my_func)   # ✅ for a Series
df.map(my_func)         # ✅ for a whole DataFrame

🎯 Mini Challenge: Order Summary

Recode a status column with map, then build a row total with a row-wise apply.

import pandas as pd

df = pd.DataFrame({
    "qty": [2, 1, 3],
    "price": [9.99, 19.5, 4.0],
    "status": ["P", "S", "P"],
})

# 1. Recode the status codes to words with map + a dict
df["status"] = df["status"].map({"P": "Paid", "S": "Shipped"})

# 2. Row-wise apply to compute line totals (axis=1)
df["line_total"] = df.apply(lambda r: round(r["qty"] * r["price"], 2), axis=1)

print(df)
# line_total: 19.98, 19.5, 12.0

# ✅ Expected output:
#    qty  price   status  line_total
# 0    2   9.99     Paid       19.98
# 1    1  19.50  Shipped       19.50
# 2    3   4.00     Paid       12.00

❓ Frequently Asked Questions

Lesson complete — you can transform data your way!

You now know when to use map for recoding, apply for flexible per-column or per-row logic, and applymap / DataFrame.map for every cell. Lambdas keep these transformations short and readable.

Practice quiz

Continue this course

• Previous: idxmax & idxmin: Find the Row Behind the Max
• Next: String Operations with .str — Clean and transform text columns with vectorised string methods
• Quick reference: Pandas cheat sheet › Transforming

Frequently asked questions

What is the difference between map and apply in pandas?

Series.map is used to substitute or transform values element by element, and it accepts a dictionary, a Series, or a function. Series.apply only accepts a function but is more flexible for richer logic. On a DataFrame, apply runs a function over whole columns or rows, which map cannot do.

What does the axis argument do in DataFrame.apply?

axis=0 (the default) passes each column to your function as a Series, so you typically get one result per column. axis=1 passes each row as a Series, so you get one result per row. Use axis=1 when a new value depends on several columns of the same row.

Is applymap removed in newer pandas versions?

DataFrame.applymap was deprecated in pandas 2.1 in favour of DataFrame.map, which does the same elementwise operation. applymap still works for now but emits a warning, so prefer DataFrame.map in new code while knowing both names refer to the same idea.