如何将熊猫数据框转换为命名元组

Question

How to convert a pandas dataframe to namedtuple? This task is going towards multiprocessing work.

def df2namedtuple(df):
   return tuple(df.row)

Answer 1

itertuples has option name and index . You may use them to return exact output as your posted function:

sample df:

df:
    A   B   C   D
0  32  70  39  66
1  89  30  31  80
2  21   5  74  63

list(df.itertuples(name='Row', index=False))

Out[1130]:
[Row(A=32, B=70, C=39, D=66),
 Row(A=89, B=30, C=31, D=80),
 Row(A=21, B=5, C=74, D=63)]

Answer 2

Answer from Dan in https://groups.google.com/forum/#!topic/pydata/UaF6Y1LE5TI

from collections import namedtuple

def iternamedtuples(df):
    Row = namedtuple('Row', df.columns)
    for row in df.itertuples():
        yield Row(*row[1:])