Lesson 06

Creating a DataFrame

Build a DataFrame from a dict of columns, a list of rows, a list of dicts, or a dict of Series, and set the index and column order yourself.

Intermediate25 min

Creating a DataFrame hero Module 24 Lesson 6

A DataFrame is a table: a row index, named columns, and the values. You can build one from a dict where each key is a column, or from rows where each row is a list, a tuple, or a dict.

A four-column DataFrame with the first column highlighted

From a dict of columns

This is the usual way. Each key is a column name and each value is the column data. All the lists must be the same length.

python

import pandas as pd

team = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey"],
    "hours": [42, 31, 55],
    "score": [88, 74, 91],
})
print(team)
print(team.columns.tolist())

Output:

text

     name  hours  score
0   Avery     42     88
1  Jordan     31     74
2   Casey     55     91
['name', 'hours', 'score']

From rows

A list of tuples or lists is one row each. Pass columns to name them, or you get 0, 1, 2 as headings. index names the rows.

python

import pandas as pd

rows = [("Avery", 42, 88), ("Jordan", 31, 74), ("Casey", 55, 91)]
team = pd.DataFrame(rows, columns=["name", "hours", "score"], index=["a", "b", "c"])
print(team)

print(pd.DataFrame(rows))

Output:

text

     name  hours  score
a   Avery     42     88
b  Jordan     31     74
c   Casey     55     91
        0   1   2
0   Avery  42  88
1  Jordan  31  74
2   Casey  55  91

From a list of dicts

Each dict is one row and its keys are the column names. A key that is missing from one row becomes NaN, which makes this shape forgiving when records differ.

python

import pandas as pd

records = [
    {"name": "Avery", "hours": 42, "score": 88},
    {"name": "Jordan", "hours": 31},
    {"name": "Casey", "hours": 55, "score": 91, "city": "Northgate"},
]
print(pd.DataFrame(records))

Output:

text

     name  hours  score       city
0   Avery     42   88.0        NaN
1  Jordan     31    NaN        NaN
2   Casey     55   91.0  Northgate

From a dict of Series

When the values are Series, pandas lines the columns up on their labels rather than on their order. Any label that is missing from a column is filled with NaN.

python

import pandas as pd

hours = pd.Series({"Avery": 42, "Jordan": 31, "Casey": 55})
score = pd.Series({"Casey": 91, "Avery": 88})
print(pd.DataFrame({"hours": hours, "score": score}))

Output:

text

        hours  score
Avery      42   88.0
Casey      55   91.0
Jordan     31    NaN

Choose the index and the columns

columns can also pick and reorder. set_index promotes a column to the row index, and reset_index puts it back as a normal column.

python

import pandas as pd

team = pd.DataFrame({
    "name": ["Avery", "Jordan", "Casey"],
    "hours": [42, 31, 55],
    "score": [88, 74, 91],
})
print(team[["score", "name"]])

by_name = team.set_index("name")
print(by_name)
print(by_name.reset_index())

Output:

text

   score    name
0     88   Avery
1     74  Jordan
2     91   Casey
        hours  score
name                
Avery      42     88
Jordan     31     74
Casey      55     91
     name  hours  score
0   Avery     42     88
1  Jordan     31     74
2   Casey     55     91

Key takeaways

A dict of lists is one column per key
A list of tuples is one row each, and columns names them
A list of dicts fills any missing key with NaN
A dict of Series lines up on labels, not on order
Next: DataFrame Attributes
Previous: Series Statistics

Loading playground…

Check your understanding

Creating a DataFrame Quiz

In pd.DataFrame({"a": [1, 2]}), what is "a"?

1 / 10

← Series Statistics

DataFrame Attributes →