使用pandas read_csv方法进行Python多级索引

Question

I want to read the following table as a pandas dataframe

Say the dataframe is df, the purpose is to query df['acct_id']['A']['0-3_mon] should give me 10

I have done it for panel data, where everything is a column and then you create a multi-level-index for both cross-section and time-series.

But over here, the source data itself has more than two levels of columns. How do I read this csv as a multi-level index? I am stuck here, any idea.

Some of the similar work if you want to look at - https://lectures.quantecon.org/py/pandas_panel.html

Thanks a lot.

Answer 1

Create DataFrame with MultiIndex , because deprecate panel :

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

And then select by slicers :

idx = pd.IndexSlice
print (df.loc[1, idx['A', '0-3_mon']])

Sample : with no Multindex names:

import pandas as pd

temp=u"""A;A;B;B
0-3_mon;3-6_mon;0-3_mon;3-6_mon
1;10;12;14;18
2;11;15;17;19
3;13;16;21;20"""
#after testing replace 'pd.compat.StringIO(temp)' to 'filename.csv'
df = pd.read_csv(pd.compat.StringIO(temp), sep=";", header=[0,1])
print (df)
        A               B        
  0-3_mon 3-6_mon 0-3_mon 3-6_mon
1      10      12      14      18
2      11      15      17      19
3      13      16      21      20

print (df.columns)
MultiIndex(levels=[['A', 'B'], ['0-3_mon', '3-6_mon']],
           labels=[[0, 0, 1, 1], [0, 1, 0, 1]])

idx = pd.IndexSlice
print (df.loc[1, idx['A', '0-3_mon']])
10

Sample with specified names of MultiIndex:

import pandas as pd

temp=u"""acct_id;A;A;B;B
level;0-3_mon;3-6_mon;0-3_mon;3-6_mon
1;10;12;14;18
2;11;15;17;19
3;13;16;21;20"""
#after testing replace 'pd.compat.StringIO(temp)' to 'filename.csv'
df = pd.read_csv(pd.compat.StringIO(temp), sep=";", index_col=[0], header=[0,1])
print (df)
acct_id       A               B        
level   0-3_mon 3-6_mon 0-3_mon 3-6_mon
1            10      12      14      18
2            11      15      17      19
3            13      16      21      20

print (df.columns)

MultiIndex(levels=[['A', 'B'], ['0-3_mon', '3-6_mon']],
           labels=[[0, 0, 1, 1], [0, 1, 0, 1]],
           names=['acct_id', 'level'])

idx = pd.IndexSlice
print (df.loc[1, idx['A', '0-3_mon']])
10