Turn pandas columns to a Python dictionary

Create a dictionary from pandas columns

We are able to convert one or more pandas Series to a dictionary using the to_dict Series and DataFrame methods:

#Series
my_series.to_dict()

#Multiple DataFrame columns
my_df[[col1, col2]].to_dict()

Step #1 Create DataFrame

We’ll start by importing the pandas Data Analysis library and creating a DataFrame.

import pandas pd

raw_dict = {
             'channel' : ['B2C', 'B2B', 'Online'],
             'sales' : [250, 345, 432]
            }

perf_df = pd.DataFrame (raw_dict)

perf_df.head()

Here’s our dataset:

channelsales
0B2C250
1B2B345
2Online432

Step #2: Using to_dict to convert a single column to a dictionary object

We can easily pull the contents of a Series into a dictionary:

perf_df.channel.to_dict()

This will return a dictionary made of key / value pairs consisting of the Series index and elements:

{0: 'B2C', 1: 'B2B', 2: 'Online'}

Step #3: convert multiple column to dictionary

Next case is to convert the entire DataFrame to a Python dict object:

sales_dict = perf_df.to_dict()
print(sales_dict)

Or converting selected multiple columns, by passing a list of columns to the to_dict() function

sales_dict = perf_df [['channel', 'sales']].to_dict()
print(sales_dict)

Both will return the following object:

{'channel': {0: 'B2C', 1: 'B2B', 2: 'Online'}, 'sales': {0: 250, 1: 345, 2: 432}}

Step #4: Zip two columns and make a dictionary

Another way to turn our columns to a dictionary object is to zip the two Series, then convert them into a list and last build the dictionary.

sales_lst = list(zip(perf_df.channel,perf_df.sales ))
sales_dict = dict(sales_lst)

This will result in:

{'sales': {'B2C': 250, 'B2B': 345, 'Online': 432}}

Note that the dictionary doesn’t include the DataFrame index column. A similar way to accomplish the same result would have been:

sales_dict = perf_df.set_index('channel').to_dict()

Related learning:

How to convert a groupby object to a pandas DataFrame?