Python Numpy：如何将带有一对逗号分隔浮点数的文本文件转换为多维“ndarray”

Question

I am new to numpy and i need to convert text file with data

219062.60893,395935.54879 219332.52719,395961.82402 219301.47465,395688.32278 219036.33371,395677.57382 218761.63814,395494.84155 219164.12686,395438.70811 219086.49551,395244.03255 218758.05515,395308.52630

to a numpy ndarray of

[[[219062.60893,395935.54879],[219332.52719,395961.82402],[219301.47465,395688.32278],[219036.33371,395677.57382]], [[218761.63814,395494.84155],[219164.12686,395438.70811],[219086.49551,395244.03255],[218758.05515,395308.52630]]]

what i tried is this

 textLineArray = np.loadtxt(filePath, str, None, None, None, 0, None, False,0,'bytes',None)

gives me

[['219062.60893,395935.54879' '219332.52719,395961.82402'
 '219301.47465,395688.32278' '219036.33371,395677.57382'],
['218761.63814,395494.84155' '219164.12686,395438.70811'
 '219086.49551,395244.03255' '218758.05515,395308.52630']]

and after further spiting with space

spaceTextLineArray = np.char.split(textLineArray, ' ', maxsplit=None)

I get this

[[list(['219062.60893,395935.54879']) list(['219332.52719,395961.82402'])
 list(['219301.47465,395688.32278']) list(['219036.33371,395677.57382'])],[list(['218761.63814,395494.84155']) list(['219164.12686,395438.70811'])
 list(['219086.49551,395244.03255']) list(['218758.05515,395308.52630'])]]

quite their but not exactly their don't know how to get-rid of single quotes

Answer 1

First solution

Try this code:

import numpy as np

data = []
with open('data.txt') as my_file:      
    for line in my_file:  
        data.append([list(map(float ,x.split(','))) for x in line.split(' ')])
arr_data = np.array(data)

and arr_data will contain your numpy array:

array([[[219062.60893, 395935.54879],
        [219332.52719, 395961.82402],
        [219301.47465, 395688.32278],
        [219036.33371, 395677.57382]],

       [[218761.63814, 395494.84155],
        [219164.12686, 395438.70811],
        [219086.49551, 395244.03255],
        [218758.05515, 395308.5263 ]]])

Brief explanation:

1. Read your file line by line
2. Format and store one line data in a list
3. Convert list to numpy array

---

Second solution

Another solution, without external for-loops, which produces the same results:

arr_data = [[list(map(float, a.split(','))) for a in s] for s in np.loadtxt('myData.csv', dtype=str)]

Comparison of Execution Times

I used a file like yours format, with 5000 lines, and the results obtained are the following:

• First Solution:

  # 41.4 ms ± 4.08 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

• Second Solution:

  # 84.6 ms ± 6.06 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

The first solution I proposed to you seems to be about twice as fast.

---

Extra

If instead you have a standard csv format , and you want to upload them directly to a numpy array, you can do so:

from numpy import genfromtxt
arr_data = genfromtxt('file_data.csv',delimiter=',')

and my_data will contain your numpy array.