How to make a timestamp from two elements of an array in python?

Question

I have a data file in the form of

Day      Time   Value
20100511,001012,8
20100511,001052,14

I load the file in a SciPy script with

data = np.loadtxt("data.txt", delimiter=",", usecols=(0,1,2))
y= data[:,2]

How can I make x = timestamp from the data given in the first two columns?

Answer 1

使用Pandas很容易。首先，解析CSV：

scipy.optimize.leastsq

然后：

data = pd.read_csv('data.txt', skiprows=1, names=['Day', 'Time', 'Value'], dtype=str)

这给了你：

pd.to_datetime(data.Day + 'T' + data.Time) # ISO 8601 compact format

或者，发烧友，一步到位：

0   2010-05-11 00:10:12
1   2010-05-11 00:10:52
dtype: datetime64[ns]

Answer 2

假设您不想使用pandas，可以使用numpy.apply_along_axis：

In [77]: np.apply_along_axis(lambda x: datetime.strptime(x[0][2:-1] + x[1][2:-1], "%Y%m%d%H%M%S"), 1, data[1:])
Out[77]: 
array([datetime.datetime(2010, 5, 11, 0, 10, 12),
       datetime.datetime(2010, 5, 11, 0, 10, 52)], dtype=object)

笨拙的字符串切片，因为np.read_txt在开头添加了额外的b'，在结尾添加了'。