Pandas read_csv import results in error

Question

My csv is as follows (MQM Q.csv):

Date-Time,Value,Grade,Approval,Interpolation Code 
31/08/2012 12:15:00,,41,1,1 
31/08/2012 12:30:00,,41,1,1 
31/08/2012 12:45:00,,41,1,1 
31/08/2012 13:00:00,,41,1,1 
31/08/2012 13:15:00,,41,1,1 
31/08/2012 13:30:00,,41,1,1 
31/08/2012 13:45:00,,41,1,1 
31/08/2012 14:00:00,,41,1,1 
31/08/2012 14:15:00,,41,1,1

The first few lines have no "Value" entries but they start later on.

Here is my code:

import pandas as pd 
from StringIO import StringIO
Q = pd.read_csv(StringIO("""/cygdrive/c/temp/MQM Q.csv"""), header=0, usecols=["Date-Time", "Value"], parse_dates=True, dayfirst=True, index_col=0)

I get the following error:

Traceback (most recent call last):
  File "daily.py", line 4, in <module>
    Q = pd.read_csv(StringIO("""/cygdrive/c/temp/MQM Q.csv"""), header=0, usecols=["Date-Time", "Value"], parse_dates=True, dayfirst=True, index_col=0)
  File "/usr/lib/python2.7/site-packages/pandas-0.14.0-py2.7-cygwin-1.7.30-x86_64.egg/pandas/io/parsers.py", line 443, in parser_f
    return _read(filepath_or_buffer, kwds)
  File "/usr/lib/python2.7/site-packages/pandas-0.14.0-py2.7-cygwin-1.7.30-x86_64.egg/pandas/io/parsers.py", line 228, in _read
    parser = TextFileReader(filepath_or_buffer, **kwds)
  File "/usr/lib/python2.7/site-packages/pandas-0.14.0-py2.7-cygwin-1.7.30-x86_64.egg/pandas/io/parsers.py", line 533, in __init__
    self._make_engine(self.engine)
  File "/usr/lib/python2.7/site-packages/pandas-0.14.0-py2.7-cygwin-1.7.30-x86_64.egg/pandas/io/parsers.py", line 670, in _make_engine
    self._engine = CParserWrapper(self.f, **self.options)
  File "/usr/lib/python2.7/site-packages/pandas-0.14.0-py2.7-cygwin-1.7.30-x86_64.egg/pandas/io/parsers.py", line 1067, in __init__
    col_indices.append(self.names.index(u))
ValueError: 'Value' is not in list

Answer 1

This appears to be a bug with the csv parser, firstly this works:

df = pd.read_csv('MQM Q.csv')

also this works:

df = pd.read_csv('MQM Q.csv', usecols=['Value'])

but if I want Date-Time then it fails with the same error message as yours.

So I noticed it was utf-8 encoded and so I converted using notepad++ to ANSI and it worked, I then tried utf-8 without BOM and it also worked.

I then converted it to utf-8 (presumably there is now a BOM) and it failed with the same error as before, so I don't think you are imaging this now and this looks like a bug.

I am using python 3.3, pandas 0.14 and numpy 1.8.1

To get around this do this:

df = pd.read_csv('MQM Q.csv', usecols=[0,1], parse_dates=True, dayfirst=True, index_col=0)

This will set your index to the Date-Time column which will correctly convert to a datetimeindex.

In [40]:

df.index
Out[40]:
<class 'pandas.tseries.index.DatetimeIndex'>
[2012-08-31 12:15:00, ..., 2013-11-28 10:45:00]
Length: 43577, Freq: None, Timezone: None