Question

我想从文件中读取时间字符串和数据，但是当我使用loadtxt时，我无法同时读取字符串和数字，因为字符串不是浮点数。所以我尝试使用genfromtxt并使用delimiter=[]+[]+[]来记录我所拥有的列，但字符串的读取方式与nan类似。我想像时间数组（date2num，datetime或类似）一样直接读取时间，以便能够以正确的形式在matplotlib中绘图。那么，我该怎么办？我在下面留下一个mi列表（显然，这是更多的数据）：

GOES data for time interval: 20-Feb-2014 00:00:00.000 to 27-Feb-2014 00:00:00.000
Current time: 23-Mar-2014 21:52:00.00

Time at center of bin        1.0 - 8.0 A    0.5 - 4.0 A  Emission Meas           Temp
                              watts m^-2     watts m^-2    10^49 cm^-3             MK
20-Feb-2014 00:00:00.959     4.3439e-006    3.9946e-007        0.30841         10.793
20-Feb-2014 00:00:02.959     4.3361e-006    3.9835e-007        0.30801         10.789
20-Feb-2014 00:00:04.959     4.3413e-006    3.9501e-007        0.30994         10.743
20-Feb-2014 00:00:06.959     4.3361e-006    3.9389e-007        0.30983         10.735
20-Feb-2014 00:00:08.959     4.3361e-006    3.9278e-007        0.31029         10.722
20-Feb-2014 00:00:10.959     4.3387e-006    3.9278e-007        0.31058         10.719
20-Feb-2014 00:00:12.959     4.3361e-006    3.9278e-007        0.31029         10.722
20-Feb-2014 00:00:14.959     4.3361e-006    3.9055e-007        0.31122         10.695
20-Feb-2014 00:00:16.959     4.3334e-006    3.8721e-007        0.31234         10.657

根据建议，我使用以下方式阅读数据：

pd.read_csv('/filename',sep='\s\s+',header=5,
               names=['time','band1','band2','emeas','temp'])

我读了数据，但只是一个问题，当我打印数据时出现：

                       time     band1  band2    emeas    temp
0  20-Feb-2014 00:00:03.005  0.000004      0  0.31000  10.866
1  20-Feb-2014 00:00:05.052  0.000004      0  0.31199  10.819
2  20-Feb-2014 00:00:07.102  0.000004      0  0.31190  10.811
3  20-Feb-2014 00:00:09.149  0.000004      0  0.31237  10.798
4  20-Feb-2014 00:00:11.199  0.000004      0  0.31266  10.795
5  20-Feb-2014 00:00:13.245  0.000004      0  0.31237  10.798
6  20-Feb-2014 00:00:15.292  0.000004      0  0.31334  10.770
7  20-Feb-2014 00:00:17.342  0.000004      0  0.31451  10.732
8  20-Feb-2014 00:00:19.389  0.000004      0  0.31451  10.732
9  20-Feb-2014 00:00:21.439  0.000004      0  0.31421  10.735

所以，显然band1和band2的数据已经四舍五入。实际上，当绘图时它似乎是正确的（非圆形），但为什么在框架中看起来像那样。

Answer 1

使用正则表达式可能有更优雅的解决方案，但这也有效。

from datetime import datetime

input_file = open("path/filename")
for line in input_file:
    line_parts = line.split()
    if len(line_parts) > 1:
        try:
            # This is now a datetime object
            timestamp = datetime.strptime(line_parts[0] + " " + line_parts[1], "%d-%b-%Y %H:%M:%S.%f")
            # Do stuff with data here (each stored seperately in line_parts list)
            # For instance printing everything.
            print("DateTime Object: " + str(timestamp))
            print("Data: " + str(line_parts[2:]))

            # Cast data to floats for use in arithmetic
            data_point_one = float(line_parts[2])
            print ("data_point_one * 2 = " + str(data_point_one * 2))

        except ValueError:
            # Lines that don't start with a timestamp take this route...
            continue

Answer 2

您可以使用pandas.read_csv()，因为sep参数（相当于delimiter中的numpy.genfromtxt）接受正则表达式。然后，用：

import pandas as pd

pd.read_csv('test.txt', sep='\s\s+', header=4)

您将获得所需的输出。

从文件中读取日期和数据（Python）

2 个答案: