Skip rows during csv import pandas

I'm trying to import a .csv file using pandas.read_csv(), however, I don't want to import the 2nd row of the data file (the row with index = 1 for 0-indexing).

I can't see how not to import it because the arguments used with the command seem ambiguous:

From the pandas website:

skiprows : list-like or integer

Row numbers to skip (0-indexed) or number of rows to skip (int) at the start of the file."

If I put skiprows=1 in the arguments, how does it know whether to skip the first row or skip the row with index 1?

286365 次浏览
小开
最佳答案

You can try yourself:

>>> import pandas as pd
>>> from StringIO import StringIO
>>> s = """1, 2
... 3, 4
... 5, 6"""
>>> pd.read_csv(StringIO(s), skiprows=[1], header=None)
0  1
0  1  2
1  5  6
>>> pd.read_csv(StringIO(s), skiprows=1, header=None)
0  1
0  3  4
1  5  6
小开

I don't have reputation to comment yet, but I want to add to alko answer for further reference.

From the docs:

skiprows: A collection of numbers for rows in the file to skip. Can also be an integer to skip the first n rows

小开

I got the same issue while running the skiprows while reading the csv file. I was doning skip_rows=1 this will not work

Simple example gives an idea how to use skiprows while reading csv file.

import pandas as pd


#skiprows=1 will skip first line and try to read from second line
df = pd.read_csv('my_csv_file.csv', skiprows=1)  ## pandas as pd


#print the data frame
df
小开

skip[1] will skip second line, not the first one.

小开

All of these answers miss one important point -- the n'th line is the n'th line in the file, and not the n'th row in the dataset. I have a situation where I download some antiquated stream gauge data from the USGS. The head of the dataset is commented with '#', the first line after that are the labels, next comes a line that describes the date types, and last the data itself. I never know how many comment lines there are, but I know what the first couple of rows are. Example:

> # ----------------------------- WARNING ----------------------------------
> # Some of the data that you have obtained from this U.S. Geological Survey database
> # may not have received Director's approval. ... agency_cd    site_no datetime    tz_cd   139719_00065    139719_00065_cd
> 5s    15s 20d 6s  14n 10s USGS    08041780    2018-05-06 00:00    CDT 1.98    A

It would be nice if there was a way to automatically skip the n'th row as well as the n'th line.

As a note, I was able to fix my issue with:

import pandas as pd
ds = pd.read_csv(fname, comment='#', sep='\t', header=0, parse_dates=True)
ds.drop(0, inplace=True)
小开

Indices in read_csv refer to line/row numbers in your csv file (the first line has the index 0). You have the following options to skip rows:

from io import StringIO


csv = \
"""col1,col2
1,a
2,b
3,c
4,d
"""
pd.read_csv(StringIO(csv))


# Output:
col1 col2  # index 0
0     1    a  # index 1
1     2    b  # index 2
2     3    c  # index 3
3     4    d  # index 4

Skip two lines at the start of the file (index 0 and 1). Column names are skipped as well (index 0) and the top line is used for column names. To add column names use names = ['col1', 'col2'] parameter:

pd.read_csv(StringIO(csv), skiprows=2)


# Output:
2  b
0  3  c
1  4  d

Skip second and fourth lines (index 1 and 3):

pd.read_csv(StringIO(csv), skiprows=[1, 3])


# Output:
col1 col2
0     2    b
1     4    d

Skip last two lines:

pd.read_csv(StringIO(csv), engine='python', skipfooter=2)


# Output:
col1 col2
0     1    a
1     2    b

Use a lambda function to skip every second line (index 1 and 3):

pd.read_csv(StringIO(csv), skiprows=lambda x: (x % 2) != 0)


# Output:
col1 col2
0     2    b
1     4    d

提交答案