RogerWilco77 RogerWilco77 - 10 months ago 109
Python Question

Read csv with in Python and Pandas

I am reading a csv file with German date format.
Seems like it worked ok in this post:

Picking dates from an imported CSV with pandas/python

However, it seems like in my case the date is not recognized as such.
I could not find any wrong string in the test file.

import pandas as pd
import numpy as np

%matplotlib inline
import matplotlib.pyplot as plt

from matplotlib import style
from pandas import DataFrame


df = pd.read_csv('testdata.csv', dayfirst=True, parse_dates=True)


This results in:


So, the Column with the dates is not recognized as such.
What am I doing wrong here?
Or is this date format simply not compatible?

  • OSX 10.10.3

  • Anaconda conda 3.13.0

  • Python 3.4.3-0

  • iPython notebook 3.1.0

Answer Source

If you use parse_dates=True then read_csv tries to parse the index as a date. Therefore, you would also need to declare the first column as the index with index_col=[0]:

In [216]: pd.read_csv('testdata.csv', dayfirst=True, parse_dates=True, index_col=[0])
            morgens  mittags  abends
2015-03-16      382      452     202
2015-03-17      288      467     192

Alternatively, if you don't want the Datum column to be an index, you could use parse_dates=[0] to explicitly tell read_csv to parse the first column as dates:

In [217]: pd.read_csv('testdata.csv', dayfirst=True, parse_dates=[0])
       Datum  morgens  mittags  abends
0 2015-03-16      382      452     202
1 2015-03-17      288      467     192

Under the hood read_csv uses dateutil.parser.parse to parse date strings:

In [218]: import dateutil.parser as DP

In [221]: DP.parse('16.03.2015', dayfirst=True)
Out[221]: datetime.datetime(2015, 3, 16, 0, 0)

Since dateutil.parser has no trouble parsing date strings in DD.MM.YYYY format, you don't have to declare a custom date parser here.