5 Best Ways to Retrieve the Shape of Data with Python Pandas

by

💡 Problem Formulation: When working with datasets in Python’s Pandas library, understanding the structure of your data is crucial. Often, you’ll need to know the number of rows and columns in your DataFrame or Series, which is represented as a tuple (rows, columns). This article explains how to acquire this tuple and what each method’s strengths and weaknesses are.

Method 1: Using the shape Attribute

The shape attribute of a DataFrame or Series returns the dimensionality of the data structure as a tuple. This is the most straightforward and commonly used method to understand the size and structure of your data.

Here’s an example:

import pandas as pd

# Create a simple DataFrame
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
# Get the shape of the DataFrame
print(df.shape)

Output:

(3, 2)

This code snippet creates a simple DataFrame with three rows and two columns, then prints its shape, which is (3, 2), using the shape attribute.

Method 2: Using len() and DataFrame.columns

You can also determine the number of rows and columns by combining the use of len() function on the DataFrame itself for rows, and on its columns attribute for columns.

Here’s an example:

import pandas as pd

# Create a simple DataFrame
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
# Get the number of rows and columns
shape_tuple = (len(df), len(df.columns))
print(shape_tuple)

Output:

(3, 2)

This code example uses the built-in len() function to determine the number of rows and columns in the DataFrame, then prints this information as a shape tuple (3, 2).

Method 3: Using numpy.shape

The NumPy library, which is often used alongside Pandas, provides a shape function as well, which can be applied to Pandas DataFrames and Series, returning the expected shape tuple.

Here’s an example:

import pandas as pd
import numpy as np

# Create a simple DataFrame
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
# Get the shape using numpy's shape function
shape_tuple = np.shape(df)
print(shape_tuple)

Output:

(3, 2)

In this snippet, the NumPy library’s shape function is used to achieve the same result as the DataFrame’s shape attribute, confirming the data’s structure is 3 rows by 2 columns.

Method 4: Using DataFrame size and len()

You can calculate the number of rows by dividing the total number of elements in the DataFrame (given by the size attribute) by the number of columns. This is a less direct approach and is mainly of theoretical interest.

Here’s an example:

import pandas as pd

# Create a simple DataFrame
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
# Calculate the number of rows
rows = df.size // len(df.columns)
# Create the shape tuple
shape_tuple = (rows, len(df.columns))
print(shape_tuple)

Output:

(3, 2)

This snippet uses the DataFrame’s size attribute to find the total number of elements and then calculates the number of rows by dividing this by the number of columns. The shape of the DataFrame is then printed as a tuple.

Bonus One-Liner Method 5: Using a List Comprehension

For those who love one-liners, you can use a list comprehension with len() function applied to both the DataFrame and its columns attribute, all in a single expressive line.

Here’s an example:

import pandas as pd

# Create a simple DataFrame
df = pd.DataFrame({'A': [1, 2, 3], 'B': [4, 5, 6]})
# Get the shape as a tuple using list comprehension
shape_tuple = tuple([len(df), len(df.columns)])
print(shape_tuple)

Output:

(3, 2)

By this elegant approach, a shape tuple is produced through a compact and readable one-liner that applies the len() function twice within a list comprehension.

Summary/Discussion

  • Method 1: shape Attribute. The most straightforward and recommended way to get the shape of a DataFrame or Series. It’s concise and idiomatic to Pandas. The downside is the lack of versatility in case of more complex data manipulations.
  • Method 2: len() and DataFrame.columns. Offers a Python-native way to find the data shape without relying on Pandas methods. It’s easy to understand but more verbose than using the shape attribute.
  • Method 3: NumPy’s shape Function. Useful when working with NumPy arrays and when NumPy is already a dependency in the project. It’s as straightforward as the Pandas shape attribute.
  • Method 4: size and len(). This is more of an indirect calculation of the shape. It could potentially be useful in specific cases, although it’s generally more cumbersome than necessary.
  • Bonus Method 5: List Comprehension. A concise one-liner for enthusiasts who value brevity. However, this method may not be as immediately clear to beginners or as readable as explicitly stating shape.