5 Best Ways to Create a Pandas DataFrame from an Array

by

💡 Problem Formulation: Converting an array into a DataFrame is a common task in data analysis. This involves taking input like a NumPy array or a list of lists and transforming it into a structured DataFrame using the Pandas library. The expected output is a Pandas DataFrame with rows and columns that reflect the structure and data of the original array.

Method 1: Using DataFrame Constructor

The Pandas DataFrame constructor is the most straightforward method to create a DataFrame from an array. You simply pass the array directly into the constructor, and optionally specify column names if required. The result is a neatly formatted DataFrame that presents the array in a tabular fashion.

Here’s an example:

import pandas as pd
import numpy as np

data_array = np.array([[1, 2, 3], [4, 5, 6]])
df = pd.DataFrame(data_array, columns=['A', 'B', 'C'])

print(df)

The output of this code snippet:

   A  B  C
0  1  2  3
1  4  5  6

This code imports Pandas and NumPy, creates a 2-dimensional NumPy array, and then passes this array to the Pandas DataFrame constructor, resulting in a DataFrame with three columns labeled ‘A’, ‘B’, and ‘C’.

Method 2: DataFrame from List of Lists

When dealing with a plain Python list of lists, you can also use the DataFrame constructor by simply passing the list directly. This is useful when you’re not working with NumPy and your data is already in a list format.

Here’s an example:

import pandas as pd

data_list = [[7, 8, 9], [10, 11, 12]]
df = pd.DataFrame(data_list, columns=['X', 'Y', 'Z'])

print(df)

The output of this code snippet:

    X   Y   Z
0   7   8   9
1  10  11  12

This snippet creates a DataFrame from a list of lists by passing the list object to the DataFrame constructor. This results in a DataFrame with the same number of columns as there are elements in the sublists, and as many rows as there are sublists.

Method 3: DataFrame with Custom Indices

Adding indices is an important feature when you want to identify rows with specific labels rather than default integer-based indices. By using the index argument in the DataFrame constructor, you can assign a custom index to your data.

Here’s an example:

import pandas as pd

array = [[13, 14], [15, 16]]
df = pd.DataFrame(array, columns=['Column1', 'Column2'], index=['Row1', 'Row2'])

print(df)

The output of this code snippet:

      Column1  Column2
Row1       13       14
Row2       15       16

This code creates a DataFrame with custom row indices ‘Row1’ and ‘Row2’. The DataFrame is populated with the data from the 2-dimensional array along with the specified column names.

Method 4: Using pd.DataFrame.from_records()

The from_records() method is particularly useful when dealing with a list of tuples or arrays. It assumes each tuple or array in the list is a record, and the resulting DataFrame uses these records as rows.

Here’s an example:

import pandas as pd

records = [(20, 'red'), (21, 'blue')]
df = pd.DataFrame.from_records(records, columns=['Number', 'Color'])

print(df)

The output of this code snippet:

   Number Color
0      20   red
1      21  blue

This example takes a list of tuples, with each tuple representing a record, and uses pd.DataFrame.from_records() to create a DataFrame with columns ‘Number’ and ‘Color’.

Bonus One-Liner Method 5: Using pd.DataFrame() with Zip

For a quick and efficient one-liner, you can use zip to merge multiple lists into a DataFrame. This method is best when your data is already separated into individual column lists.

Here’s an example:

import pandas as pd

col1 = [22, 23]
col2 = ['green', 'yellow']
df = pd.DataFrame(list(zip(col1, col2)), columns=['Number', 'Color'])

print(df)

The output of this code snippet:

   Number   Color
0      22   green
1      23  yellow

This snippet zips two lists together into a list of tuples and instantly converts it into a DataFrame with the specified column names, all in a single line of code.

Summary/Discussion

  • Method 1: DataFrame Constructor. Simple and straightforward. Ideally used with NumPy arrays.
  • Method 2: DataFrame from List of Lists. Perfect for Python lists, maintaining simplicity without needing NumPy.
  • Method 3: DataFrame with Custom Indices. Adds the ability to label rows with custom indices. Useful for labeled data.
  • Method 4: Using pd.DataFrame.from_records(). Great for lists of tuples or records. Each tuple naturally becomes a row.
  • Bonus Method 5: Using pd.DataFrame() with Zip. Efficient one-liner for combining multiple lists into a DataFrame.