Remove leading comma in header when using pandas to_csv

Question

By default to_csv writes a CSV like

,a,b,c
0,0.0,0.0,0.0
1,0.0,0.0,0.0
2,0.0,0.0,0.0

But I want it to write like this:

a,b,c
0,0.0,0.0,0.0
1,0.0,0.0,0.0
2,0.0,0.0,0.0

How do I achieve this? I can’t set index=False because I want to preserve the index. I just want to remove the leading comma.

df = pd.DataFrame(np.zeros((3,3)), columns = ['a','b','c'])
df.to_csv("test.csv") # this results in the first example above.

Asked By: JacksonCounty

||

Source

Answer 1

Simply set a name for your index: df.index.name = 'blah'. This name will appear as the first name in the headers.

import numpy as np
import pandas as pd

df = pd.DataFrame(np.zeros((3,3)), columns = ['a','b','c'])
df.index.name = 'my_index'
print(df.to_csv())

yields

my_index,a,b,c
0,0.0,0.0,0.0
1,0.0,0.0,0.0
2,0.0,0.0,0.0

However if (as per your comment) you wish to have 3 coma-separated names in the headers while there are 4 coma-separated values in the rows of the csv, you’ll have to handcraft it. It will NOT be compliant with any csv standard format though.

Answered By: smarie

Answer 2

Alternatively, try reseting the index so it becomes a column in data frame, named index. This works with multiple indexes as well.

df = df.reset_index()
df.to_csv('output.csv', index = False)

Answered By: Parfait

Answer 3

It is possible by write only columns without index first and then data without header in append mode:

df = pd.DataFrame(np.zeros((3,3)), columns = ['a','b','c'], index=list('XYZ'))

pd.DataFrame(columns=df.columns).to_csv("test.csv", index=False)
#alternative for empty df
#df.iloc[:0].to_csv("test.csv", index=False)
df.to_csv("test.csv", header=None, mode='a')

df = pd.read_csv("test.csv")
print (df)
     a    b    c
X  0.0  0.0  0.0
Y  0.0  0.0  0.0
Z  0.0  0.0  0.0

Answered By: jezrael

Remove leading comma in header when using pandas to_csv

Question:

Answers: