CSV 数据到 Numpy 结构化数组？

Question

Name Class Species
a     1      3
b     2      4
c     3      2
a     1      3
b     2      1
c     3      2

This above mentioned data will be from CSV file. need to convert this to structured array using numpy. need header from the csv become the columns labels for the array.

Need to print the mean occurrences of each names in each class (the mean of each species for class 1, class 2, and class 3)

I used numpy.genfromtxt() .

Answer 1

This is one way to create a numpy structured array from a csv file:

import pandas as pd

arr = pd.read_csv('file.csv').to_records(index=False)

# rec.array([('a', 1, 3), ('b', 2, 4), ('c', 3, 2), ('a', 1, 3), ('b', 2, 1),
#            ('c', 3, 2)], 
#           dtype=[('Name', 'O'), ('Class', '<i8'), ('Numbers', '<i8')])

You can then work with numpy or (easier) pandas to perform your calculations.

Answer 2

Using latest numpy (1.14) on Py3.

Your sample, cleaned up:

In [93]: txt = """Name --- Class --- Numbers
    ...: a    ---------- 1    -------- 3
    ...: b    ---------- 2    -------- 4
    ...: c    ---------- 3    -------- 2
    ...: a    ---------- 1    -------- 3
    ...: b    ---------- 2     ------- 1
    ...: c    ---------- 3   --------- 2"""
In [94]: data = np.genfromtxt(txt.splitlines(), dtype=None, names=True, encoding=None)
In [95]: data
Out[95]: 
array([('a', '----------', 1, '--------', 3),
       ('b', '----------', 2, '--------', 4),
       ('c', '----------', 3, '--------', 2),
       ('a', '----------', 1, '--------', 3),
       ('b', '----------', 2, '-------', 1),
       ('c', '----------', 3, '---------', 2)],
      dtype=[('Name', '<U1'), ('f0', '<U10'), ('Class', '<i8'), ('f1', '<U9'), ('Numbers', '<i8')])

Or skipping the dashed columns:

In [96]: data = np.genfromtxt(txt.splitlines(), dtype=None, names=True, encoding=None, usecols=[0,2,4])
In [97]: data
Out[97]: 
array([('a', 1, 3), 
       ('b', 2, 4), 
       ('c', 3, 2), 
       ('a', 1, 3), 
       ('b', 2, 1),
       ('c', 3, 2)],
      dtype=[('Name', '<U1'), ('Class', '<i8'), ('Numbers', '<i8')])