How to Create a Pandas DataFrame from a Dictionary

I often find that most of my data processing starts with simple Python dictionaries.

Over the years, I have found that moving this data into a Pandas DataFrame is the quickest way to start analyzing it.

Whether you are pulling data from a web API or a local configuration file, you need a reliable way to structure it.

In this tutorial, I will show you the various ways I create DataFrames from dictionaries using firsthand examples.

Table of Contents

Method 1: Use the Default DataFrame Constructor

The most common way I create a DataFrame is by passing a dictionary directly into the standard constructor.

I usually use this when my dictionary keys represent the column headers and the values are lists of data.

Suppose we are tracking the population of some major U.S. cities. You can use the code below:

import pandas as pd

# Creating a dictionary with U.S. City data
city_data = {
    "City": ["New York City", "Los Angeles", "Chicago", "Houston", "Phoenix"],
    "State": ["NY", "CA", "IL", "TX", "AZ"],
    "Population": [8335897, 3822238, 2665039, 2302878, 1644409]
}

# Converting the dictionary to a DataFrame
df = pd.DataFrame(city_data)

print(df)

You can see the output in the screenshot below.

Create a Pandas DataFrame from a Dictionary

In my experience, this is the cleanest method if your data is already “column-oriented.”

Pandas automatically takes the keys (“City”, “State”, “Population”) and turns them into your column names.

Method 2: Use the from_dict() Method with Columns Orientation

There are times when I prefer using the explicit from_dict() method because it feels more readable in a large script.

By default, this method works similarly to the constructor, where keys become columns.

I’ve used this frequently when dealing with real estate data, like median home prices in different states.

import pandas as pd

# Dictionary of median home prices
real_estate_dict = {
    "State": ["California", "Texas", "Florida", "New York"],
    "Median_Price": [780000, 350000, 400000, 450000]
}

# Using from_dict with default orientation (columns)
df = pd.DataFrame.from_dict(real_estate_dict)

print(df)

You can see the output in the screenshot below.

How to Create a Pandas DataFrame from a Dictionary

This method is great because it clearly states your intent to create the object from a dictionary.

Method 3: Create a DataFrame with Index Orientation

Sometimes the data I receive is structured “sideways,” where the keys are meant to be the rows rather than the columns.

I see this a lot with financial reports or company profiles, where each key is a unique identifier.

In such cases, I use the orient=’index’ parameter to flip the data correctly.

import pandas as pd

# Data structured by Company Name as the key
tech_stocks = {
    "Apple": ["AAPL", 185.92, "Technology"],
    "Microsoft": ["MSFT", 420.55, "Software"],
    "Amazon": ["AMZN", 178.15, "E-commerce"]
}

# Using orient='index' and providing custom column names
df = pd.DataFrame.from_dict(tech_stocks, orient='index', 
                            columns=['Ticker', 'Price', 'Sector'])

print(df)

You can see the output in the screenshot below.

Create Pandas DataFrame from a Dictionary

When I use this, I always make sure to pass the columns argument; otherwise, Pandas will just label the columns with numbers (0, 1, 2).

Method 4: Handle a List of Dictionaries

In my professional work, especially when working with JSON data from U.S. government databases, I often get a list of dictionaries.

Each dictionary represents a single record (a row), which is very intuitive.

I find that the standard constructor handles this “row-wise” format perfectly without any extra parameters.

import pandas as pd

# A list of dictionaries representing individual employees
employee_records = [
    {"Name": "John Doe", "Dept": "Engineering", "Location": "Austin"},
    {"Name": "Jane Smith", "Dept": "Marketing", "Location": "New York"},
    {"Name": "Robert Brown", "Dept": "Sales", "Location": "Chicago"}
]

# Pandas automatically detects the keys as columns
df = pd.DataFrame(employee_records)

print(df)

One thing I’ve noticed is that if one dictionary is missing a key, Pandas will simply insert a NaN (Not a Number) value there.

Method 5: Create a DataFrame from a Nested Dictionary

I occasionally run into nested dictionaries, which can be a bit tricky to flatten.

A common scenario is having a “key of keys,” such as sales data categorized by year and then by quarter.

If I pass a nested dictionary directly, Pandas treats the outer keys as columns and the inner keys as the index.

import pandas as pd

# Nested dictionary: Year -> {Quarter: Sales}
us_sales_data = {
    2023: {"Q1": 50000, "Q2": 62000, "Q3": 58000, "Q4": 75000},
    2024: {"Q1": 52000, "Q2": 65000, "Q3": 60000, "Q4": 80000}
}

# Converting nested dict to DataFrame
df = pd.DataFrame(us_sales_data)

print(df)

In this result, the years (2023, 2024) become the headers, which might not be what you want.

If you want the years as rows, you can simply add .T at the end of the code to transpose the DataFrame.

Method 6: Specify Data Types

When I’m working with large datasets, I’ve learned that being specific about data types (dtypes) saves memory.

You can force specific types while creating the DataFrame from your dictionary.

I find this particularly useful when dealing with ZIP codes, which should often be strings rather than integers.

import pandas as pd

# Dictionary with U.S. addresses
address_data = {
    "Name": ["Alice", "Bob"],
    "ZipCode": [90210, 60601]
}

# Specifying ZIP code as a string to preserve leading zeros if any
df = pd.DataFrame.from_dict(address_data).astype({"ZipCode": str})

print(df.dtypes)
print(df)

Applying types early in the process prevents errors during later stages of data analysis.

I hope you find this tutorial helpful! Creating DataFrames from dictionaries is a foundational skill for any Python developer.

In this guide, I’ve shown you the most reliable methods I use in my daily work to get data ready for analysis.

You may read:

Bijay Kumar

I am Bijay Kumar, a Microsoft MVP in SharePoint. Apart from SharePoint, I started working on Python, Machine learning, and artificial intelligence for the last 5 years. During this time I got expertise in various Python libraries also like Tkinter, Pandas, NumPy, Turtle, Django, Matplotlib, Tensorflow, Scipy, Scikit-Learn, etc… for various clients in the United States, Canada, the United Kingdom, Australia, New Zealand, etc. Check out my profile.

enjoysharepoint.com/