Pandas 将 csv dateint 列读取到 datetime
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/26327626/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Pandas read csv dateint columns to datetime
提问by tdiddy
I'm new to both StackOverflow and pandas. I am trying to read in a large CSV file with stock market bin data in the following format:
我是 StackOverflow 和 Pandas 的新手。我正在尝试以以下格式读取包含股票市场 bin 数据的大型 CSV 文件：
date,time,open,high,low,close,volume,splits,earnings,dividends,sym
20130625,715,49.2634,49.2634,49.2634,49.2634,156.293,1,0,0,JPM
20130625,730,49.273,49.273,49.273,49.273,208.39,1,0,0,JPM
20130625,740,49.1866,49.1866,49.1866,49.1866,224.019,1,0,0,JPM
20130625,745,49.321,49.321,49.321,49.321,208.39,1,0,0,JPM
20130625,750,49.3306,49.369,49.3306,49.369,4583.54,1,0,0,JPM
20130625,755,49.369,49.369,49.369,49.369,416.78,1,0,0,JPM
20130625,800,49.369,49.369,49.3594,49.3594,1715.05,1,0,0,JPM
20130625,805,49.369,49.369,49.3306,49.3306,1333.7,1,0,0,JPM
20130625,810,49.3306,49.3786,49.3306,49.3786,1567.09,1,0,0,JPM
I have the following code to read it into a DataFrame in Pandas
我有以下代码将其读入 Pandas 中的 DataFrame
import numpy as np
import scipy as sp
import pandas as pd
import datetime as dt
fname = 'bindat.csv'
df = pd.read_csv(fname, header=0, sep=',')
The problem is that the date and time columns are read in as int64. I would like to merge these two to a single timestamp such as: 2013-06-25 07:15:00.
问题是日期和时间列被读入为 int64。我想将这两个合并为一个时间戳，例如：2013-06-25 07:15:00。
I am struggling to even get the time read in properly using:
我什至都在努力使用以下方法正确读取时间：
df['date'] = pd.to_datetime(df['date'].astype(str))
df['time'] = pd.to_datetime(df['time'].astype(str))
The first command works to convert, but the time seems weird.
第一个命令可以转换，但时间似乎很奇怪。
df.info()
<class 'pandas.core.frame.DataFrame'>
Int64Index: 9999 entries, 0 to 9998
Data columns (total 11 columns):
date 9999 non-null datetime64[ns]
time 9999 non-null object
open 9999 non-null float64
high 9999 non-null float64
low 9999 non-null float64
close 9999 non-null float64
volume 9999 non-null float64
splits 9999 non-null float64
earnings 9999 non-null int64
dividends 9999 non-null float64
sym 9999 non-null object
dtypes: datetime64[ns](1), float64(7), int64(1), object(2)None
And then I'll want to merge into a single DatetimeIndex.
然后我想合并为一个 DatetimeIndex。
Any suggestions are greatly appreciated.
任何建议都非常感谢。
Cheers!
干杯!
回答by DSM
There are quite a few ways to do this. One way to do it during read_csvwould be to use the parse_datesand date_parserarguments, telling parse_datesto combine the date and time columns and defining an inline function to parse the dates:
有很多方法可以做到这一点。一种方法read_csv是使用parse_dates和date_parser参数，告诉parse_dates组合日期和时间列并定义内联函数来解析日期：
>>> df = pd.read_csv("bindat.csv", parse_dates=[["date", "time"]],
date_parser=lambda x: pd.to_datetime(x, format="%Y%m%d %H%M"),
index_col="date_time")
>>> df
open high low close volume splits earnings dividends sym
date_time
2013-06-25 07:15:00 49.2634 49.2634 49.2634 49.2634 156.293 1 0 0 JPM
2013-06-25 07:30:00 49.2730 49.2730 49.2730 49.2730 208.390 1 0 0 JPM
2013-06-25 07:40:00 49.1866 49.1866 49.1866 49.1866 224.019 1 0 0 JPM
2013-06-25 07:45:00 49.3210 49.3210 49.3210 49.3210 208.390 1 0 0 JPM
2013-06-25 07:50:00 49.3306 49.3690 49.3306 49.3690 4583.540 1 0 0 JPM
2013-06-25 07:55:00 49.3690 49.3690 49.3690 49.3690 416.780 1 0 0 JPM
2013-06-25 08:00:00 49.3690 49.3690 49.3594 49.3594 1715.050 1 0 0 JPM
2013-06-25 08:05:00 49.3690 49.3690 49.3306 49.3306 1333.700 1 0 0 JPM
2013-06-25 08:10:00 49.3306 49.3786 49.3306 49.3786 1567.090 1 0 0 JPM
2013-06-25 16:10:00 49.3306 49.3786 49.3306 49.3786 1567.090 1 0 0 JPM
where I added an extra row at the end to make sure that hours were behaving.
我在最后添加了一个额外的行，以确保小时的行为。
