Pandas to_csv() - Convert DataFrame to CSV

The syntax of DataFrame to_csv() function is:

def to_csv(
    self,
    path_or_buf=None,
    sep=",",
    na_rep="",
    float_format=None,
    columns=None,
    header=True,
    index=True,
    index_label=None,
    mode="w",
    encoding=None,
    compression="infer",
    quoting=None,
    quotechar='"',
    line_terminator=None,
    chunksize=None,
    date_format=None,
    doublequote=True,
    escapechar=None,
    decimal=".",
)

Some of the important parameters are:

• path_or_buf: the file object to write the CSV data. If this argument is not provided, the CSV data is returned as a string.
• sep: the delimiter for the CSV data. It should be a string of length 1, the default is a comma.
• na_rep: string representing null or missing values, default is empty string.
• columns: a sequence to specify the columns to include in the CSV output.
• header: the allowed values are boolean or a list of string, default is True. If False, the column names are not written in the output. If a list of string, it’s used to write the column names. The length of the list of string should be the same as the number of columns being written in the CSV file.
• index: if True, index is included in the CSV data. If False, the index value is not written in the CSV output.
• index_label: used to specify the column name for index.

Let’s look at some common examples of using to_csv() function to convert DataFrame to CSV data.

1. Converting DataFrame to CSV String

import pandas as pd

d1 = {'Name': ['Pankaj', 'Meghna'], 'ID': [1, 2], 'Role': ['CEO', 'CTO']}

df = pd.DataFrame(d1)

print('DataFrame:\n', df)

# default CSV
csv_data = df.to_csv()
print('\nCSV String:\n', csv_data)

Output:

DataFrame:
      Name  ID Role
0  Pankaj   1  CEO
1  Meghna   2  CTO

CSV String:
 ,Name,ID,Role
0,Pankaj,1,CEO
1,Meghna,2,CTO

2. Specifying Delimiter for the CSV Output

csv_data = df.to_csv(sep='|')
print(csv_data)

Output:

|Name|ID|Role
0|Pankaj|1|CEO
1|Meghna|2|CTO

If the specified delimiter length is not 1, TypeError: “delimiter” must be a 1-character string is raised.

3. Selecting only few columns for CSV Output

csv_data = df.to_csv(columns=['Name', 'ID'])
print(csv_data)

Output:

,Name,ID
0,Pankaj,1
1,Meghna,2

Notice that the index is not considered to be a valid column.

4. Ignoring Header Row in the CSV Output

csv_data = df.to_csv(header=False)
print(csv_data)

Output:

0,Pankaj,1,CEO
1,Meghna,2,CTO

5. Setting Custom Column Names in the CSV

csv_data = df.to_csv(header=['NAME', 'ID', 'ROLE'])
print(csv_data)

Output:

,NAME,ID,ROLE
0,Pankaj,1,CEO
1,Meghna,2,CTO

Again the index is not considered as the column of DataFrame object.

6. Skipping Index Column in CSV Output

csv_data = df.to_csv(index=False)
print(csv_data)

Output:

Name,ID,Role
Pankaj,1,CEO
Meghna,2,CTO

7. Setting Index Column Name in the CSV

csv_data = df.to_csv(index_label='Sl No.')
print(csv_data)

Output:

Sl No.,Name,ID,Role
0,Pankaj,1,CEO
1,Meghna,2,CTO

8. Converting DataFrame to CSV File

with open('csv_data.txt', 'w') as csv_file:
    df.to_csv(path_or_buf=csv_file)

We are using with statement to open the file, it takes care of closing the file when the with statement block execution is finished. This code snippet will create a CSV file with the following data.

9. Null, NA, or Missing Data Representation in the CSV Output

import pandas as pd

d1 = {'Name': ['Pankaj', 'Meghna'], 'ID': [1, pd.NaT], 'Role': [pd.NaT, 'CTO']}
df = pd.DataFrame(d1)
print('DataFrame:\n', df)

csv_data = df.to_csv()
print('\nCSV String:\n', csv_data)

csv_data = df.to_csv(na_rep="None")
print('CSV String with Null Data Representation:\n', csv_data)

Output:

DataFrame:
      Name   ID Role
0  Pankaj    1  NaT
1  Meghna  NaT  CTO

CSV String:
 ,Name,ID,Role
0,Pankaj,1,
1,Meghna,,CTO

CSV String with Null Data Representation:
 ,Name,ID,Role
0,Pankaj,1,None
1,Meghna,None,CTO

• Pandas read_csv() – Reading CSV File to DataFrame
• Python Pandas Module Tutorial
• DataFrame to_csv() API Doc