Read multi-index on the columns from csv file

问题

I have a .csv file that looks like this:

Male, Male, Male, Female, Female
R, R, L, R, R
.86, .67, .88, .78, .81

I want to read that into a df, so that I have:

    Male        Female
    R       L   R
0   .86 .67 .88 .78 .81

I did:

df = pd.read_csv('file.csv', header=[0,1])

But headers does not cut it. Which results in

Empty DataFrame
Columns: [(Male, R), (Male, R), (Male, L), (Female, R), (Female, R)]
Index: []

Yet, the docs on headers says:

(...)Can be a list of integers that specify row
locations for a multi-index on the columns E.g. [0,1,3]

What am I doing wrong? How can I possibly make it work?

回答1:

I think the problem is that you have duplicated columns: two ( Female, R).

Not sure whether it's a bug or the duplicated columns are unacceptable. Here's a workaround for you:

First read the csv with tupleize_cols=True

In [61]: df = pd.read_csv('test.csv', header=[0, 1], skipinitialspace=True, tupleize_cols=True)

In [62]: df
Out[62]: 
   (Male, R)  (Male, R)  (Male, L)  (Female, R)  (Female, R)
0       0.67       0.67       0.88         0.81         0.81

[1 rows x 5 columns]

Then convert the type of the column from Index to MultiIndex

In [63]: df.columns = pd.MultiIndex.from_tuples(df.columns)

In [64]: df
Out[64]: 
   Male              Female      
      R     R     L       R     R
0  0.67  0.67  0.88    0.81  0.81

[1 rows x 5 columns]

回答2:

As of version 0.21 of pandas, MultiIndexes are created by default so df = pd.read_csv('file.csv', header=[0,1]) should do the job.

来源：https://stackoverflow.com/questions/21318865/read-multi-index-on-the-columns-from-csv-file

标签

csv

pandas

multi-index