Pandas核心能力9：日期时间转换、提取、筛选

In [1]:

import numpy as npimport pandas as pddatetime_index = pd.date_range('2019/01/18', periods = 9, freq = 'T')df = pd.DataFrame(np.random.randn(9, 2), index = datetime_index, columns = list('ab') )df.reset_index(inplace = True)df.columns = list('dab')df.dtypes

Out[1]:

d datetime64[ns]a float64b float64dtype: object

1. 提取年月日时分秒星期

In [2]:

df['years'] = df.d.dt.year # 提取年df.years.head()

Out[2]:

0 20191 20192 20193 20194 2019Name: years, dtype: int64

In [3]:

df['days'] = df.d.dt.day # 提取日df.days.head()

Out[3]:

0 181 182 183 184 18Name: days, dtype: int64

In [4]:

df['minutes'] = df.d.dt.minute # 提取分钟df.minutes.head()

Out[4]:

0 01 12 23 34 4Name: minutes, dtype: int64

In [5]:

df['weeks'] = df.d.dt.week # 提取周数df.weeks.head()

Out[5]:

0 31 32 33 34 3Name: weeks, dtype: int64

In [6]:

df['weekdays'] = df.d.dt.weekday # 提取星期,Monday=0, Sunday=6df.weekdays.head()

Out[6]:

0 41 42 43 44 4Name: weekdays, dtype: int64

2. 年月日时分秒多列合成为一列

In [7]:

df[['years', 'days', 'minutes', 'weeks', 'weekdays']] = df[['years', 'days', 'minutes', 'weeks', 'weekdays']].astype(str) # 转换列数据类型

In [8]:

df['y_d_m'] = df['years'] + '-' + df['days'] + '-' + df['minutes'] # 多列合成一列：年，日，分钟列合成为y_d_m一列df['y_d_m'].head()

Out[8]:

0 2019-18-01 2019-18-12 2019-18-23 2019-18-34 2019-18-4Name: y_d_m, dtype: object

3. 日期格式转字符串

In [9]:

from datetime import datetimedf['d_str'] = df.d.apply(lambda x: datetime.strftime(x, '%d-%m-%Y')) # 将日期烦死的d列转换为字符串的d_str列df.d_str.head()

Out[9]:

0 18-01-20191 18-01-20192 18-01-20193 18-01-20194 18-01-2019Name: d_str, dtype: object

4. 字符串转日期格式

In [10]:

df['str_datetime'] = pd.to_datetime(df.d_str) # 将字符串类型的d_str转换成日期类型的'str_datetime'。df.str_datetime.head()

Out[10]:

0 2019-01-181 2019-01-182 2019-01-183 2019-01-184 2019-01-18Name: str_datetime, dtype: datetime64[ns]

5. 日期时间筛选

In [11]:

df.loc[df.d > df.at[5, 'd']] # 筛选d列中大于第6个日期的所有数据

Out[11]:

dabyearsdaysminutesweeksweekdaysy_d_md_strstr_datetime62019-01-18 00:06:000.147799-2.7219672019186342019-18-618-01-20192019-01-1872019-01-18 00:07:001.282574-0.7348472019187342019-18-718-01-20192019-01-1882019-01-18 00:08:000.132462-0.2065542019188342019-18-818-01-20192019-01-18

In [12]:

from datetime import datetimedf.loc[df.d > datetime(2019, 1, 18, 0, 5, 0)] # 与上面结果一样

Out[12]:

dabyearsdaysminutesweeksweekdaysy_d_md_strstr_datetime62019-01-18 00:06:000.147799-2.7219672019186342019-18-618-01-20192019-01-1872019-01-18 00:07:001.282574-0.7348472019187342019-18-718-01-20192019-01-1882019-01-18 00:08:000.132462-0.2065542019188342019-18-818-01-20192019-01-18

pandas 日期时间