Lesson 08

Selecting Columns and Rows

Pull one column or several out of a DataFrame with square brackets, slice rows by position, and learn why chained selection is a poor place to assign.

Intermediate25 min

Selecting Columns and Rows hero Module 24 Lesson 8

Square brackets on a DataFrame do two different jobs. Give them a column name and you get that column. Give them a slice and you get rows. Knowing which one you asked for is most of the skill.

A table with two of its four columns picked out

One column

A single name returns a Series. The same name inside a list returns a one-column DataFrame instead. The values are the same; the shape is not, and later code cares about the difference.

python

import pandas as pd

learners = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey"],
    "hours": [42, 31, 55],
})
print(learners["name"])
print(type(learners["name"]))
print(learners[["name"]])
print(type(learners[["name"]]))

Output:

text

0     Avery
1    Jordan
2     Casey
Name: name, dtype: object
<class 'pandas.core.series.Series'>
     name
0   Avery
1  Jordan
2   Casey
<class 'pandas.core.frame.DataFrame'>

Several columns

Pass a list of names to keep several columns. The list also sets the order, so this is how you rearrange a table for a report. Asking for a name that does not exist raises a KeyError.

python

import pandas as pd

learners = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey", "Riley", "Morgan", "Quinn"],
    "track": ["Python", "NumPy", "Python", "pandas", "NumPy", "pandas"],
    "city": ["Northgate", "Eastport", "Northgate", "Westfield", "Eastport", "Westfield"],
    "hours": [42, 31, 55, 24, 47, 38],
    "score": [88, 74, 91, 65, 82, 79],
})
print(learners[["name", "score", "hours"]])
print(learners.columns.tolist())

Output:

text

     name  score  hours
0   Avery     88     42
1  Jordan     74     31
2   Casey     91     55
3   Riley     65     24
4  Morgan     82     47
5   Quinn     79     38
['name', 'track', 'city', 'hours', 'score']

Rows by position

A slice inside the brackets selects rows, not columns. It counts positions, and like every Python slice it stops before the second number. A step works too.

python

import pandas as pd

learners = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey", "Riley", "Morgan", "Quinn"],
    "track": ["Python", "NumPy", "Python", "pandas", "NumPy", "pandas"],
    "city": ["Northgate", "Eastport", "Northgate", "Westfield", "Eastport", "Westfield"],
    "hours": [42, 31, 55, 24, 47, 38],
    "score": [88, 74, 91, 65, 82, 79],
})
print(learners[0:3])
print(learners[::2])
print(learners[-2:])

Output:

text

     name   track       city  hours  score
0   Avery  Python  Northgate     42     88
1  Jordan   NumPy   Eastport     31     74
2   Casey  Python  Northgate     55     91
     name   track       city  hours  score
0   Avery  Python  Northgate     42     88
2   Casey  Python  Northgate     55     91
4  Morgan   NumPy   Eastport     47     82
     name   track       city  hours  score
4  Morgan   NumPy   Eastport     47     82
5   Quinn  pandas  Westfield     38     79

Columns and rows together

Select the columns, then slice the rows of the result. Each step is plain and easy to read, which matters more than saving a line.

python

import pandas as pd

learners = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey", "Riley", "Morgan", "Quinn"],
    "track": ["Python", "NumPy", "Python", "pandas", "NumPy", "pandas"],
    "city": ["Northgate", "Eastport", "Northgate", "Westfield", "Eastport", "Westfield"],
    "hours": [42, 31, 55, 24, 47, 38],
    "score": [88, 74, 91, 65, 82, 79],
})
picked = learners[["name", "track", "score"]]
print(picked[1:4])
print(learners[2:5][["name", "hours"]])

Output:

text

     name   track  score
1  Jordan   NumPy     74
2   Casey  Python     91
3   Riley  pandas     65
     name  hours
2   Casey     55
3   Riley     24
4  Morgan     47

Do not assign through two brackets

Reading through two selections is fine. Writing through them is not: the first selection may hand back a copy, so your change lands on the copy and the original keeps its old value. Use .loc for any write, which the next lesson covers.

python

import pandas as pd

learners = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey"],
    "score": [88, 74, 91],
})
learners.loc[learners["name"] == "Jordan", "score"] = 80
print(learners)

Output:

text

     name  score
0   Avery     88
1  Jordan     80
2   Casey     91

One rule keeps you safe: read with brackets, write with .loc or .iloc.

Key takeaways

df["col"] gives a Series and df[["col"]] gives a DataFrame
A list of names picks columns and sets their order
A slice in the brackets picks rows by position and stops before the end
Write with .loc, never through two sets of brackets
Next: loc and iloc
Previous: DataFrame Attributes

Loading playground…

Check your understanding

Selecting Columns and Rows Quiz

What type is stock["price"]?

1 / 10

← DataFrame Attributes

loc and iloc →