Table of Contents

import pandas as pd
import numpy as np
from sklearn.preprocessing import OneHotEncoder

時(shí)間(戳)的格式化

data=pd.read_csv(r"D:\downloads\tempo使用1.csv",encoding='gbk')
data.head()

pd.to_datetime(data['COLLECTTIME'])[0]

Timestamp('2014-10-01 00:00:00')

data['date']=pd.to_datetime(data['COLLECTTIME'],format="%Y-%m-%d")#數(shù)據(jù)格式化輸出
data.head()

提取時(shí)間特征：年、月、日、季度等

data['quarter']=data['date'].dt.quarter#提取季度
data.head()

data['month'],data['day']=data['date'].dt.month,data['date'].dt.day#提取月、天
data.head()

data['dayofweek']=data['date'].dt.dayofweek#一周內(nèi)的第幾天

data['weekofyear']=data['date'].dt.week#一年中的第幾周
data.head()

a=pd.to_datetime('2020-1-1')

判斷開(kāi)始、結(jié)束

#判斷開(kāi)始、結(jié)束
print(a.is_year_end,
a.is_year_start,
a.is_month_end,
a.is_month_start,
a.is_quarter_start,
a.is_quarter_end)

False True False True True False

data['hour']=data['date'].dt.hour
data.head()

判斷是否是一天的高峰時(shí)段

#是否是一天的高峰時(shí)段
data['day_hign']=data['hour'].apply(lambda x:0 if 0<=x<8 else 1)
data.head()

構(gòu)造時(shí)間特征

#構(gòu)造過(guò)去n天的統(tǒng)計(jì)數(shù)據(jù)
def get_statis_n_days_num(data,col,n):
    temp=pd.DataFrame()
    for i in range(n):
        temp=pd.concat([temp,data[col].shift((i+1)*24)],axis=1)
        data['avg_'+str(n)+"_days_"+col]=temp.mean(axis=1)
        data['median_'+str(n)+"days"+col]=temp.median(axis=1)
        data['max_'+str(n)+"_days_"+col]=temp.max(axis=1)
        data['min_'+str(n)+"_days_"+col]=temp.min(axis=1)
        data['std'+str(n)+"_days_"+col]=temp.std(axis=1)
    return data

get_statis_n_days_num(data,'CWXT_DB184C',n=7).tail()

d:\software\python\lib\site-packages\numpy\lib\nanfunctions.py:1114: RuntimeWarning: All-NaN slice encountered
  overwrite_input=overwrite_input)

時(shí)間差及其轉(zhuǎn)換

a=pd.to_datetime('2020-07-21 12:12:01')
b=pd.to_datetime('2020-07-20 12:12:23')
c=pd.to_datetime('2020-07-21 00:12:01')
d=pd.to_datetime('2020-07-21 01:00:01')

a-b#計(jì)算時(shí)間差

Timedelta('0 days 23:59:38')

(a-c)

Timedelta('0 days 12:00:00')

#將時(shí)間差，轉(zhuǎn)換成秒
(a-c).seconds

43200

(d-c).seconds

2880

# 跨天轉(zhuǎn)化秒時(shí)容易出錯(cuò)
(a-b).seconds

86378

def calTimesDiff(t1, t2):
    """
    計(jì)算時(shí)間戳之間的差值，單位h（同一天）
    :param t1:
    :param t2:
    :return:
    """
    # 1.先轉(zhuǎn)成秒
    # 2.再將秒轉(zhuǎn)換成小時(shí)
    t1_seconds = str(t1).split()[-1].split(":")
    t1_seconds = sum([int(i) * 60 ** (2 - num) for num, i in enumerate(t1_seconds)])

    t2_seconds = str(t2).split()[-1].split(":")
    t2_seconds = sum([int(i) * 60 ** (2 - num) for num, i in enumerate(t2_seconds)])
    return abs(round((t2_seconds - t1_seconds) / 3600, 3))

calTimesDiff(d,c)

0.8

a-b#計(jì)算時(shí)間差

Timedelta('0 days 23:59:38')

#將a-b轉(zhuǎn)換成小時(shí)

def convertTimestampToSec(a_b):
    """
    將時(shí)間差轉(zhuǎn)換成秒
    """
    a_b_str=str(a_b).split()
    day=int(a_b_str[0])
    sec=sum([int(i)*60**(2-num) for num,i in enumerate(a_b_str[-1].split(":"))])
    res=day*24*3600+sec
    return res

convertTimestampToSec(a-b)#23*3600+59*60+38

86378

時(shí)間的前進(jìn)與后退

pd.Timedelta

help(pd.Timedelta)

class Timedelta(_Timedelta)
 |  Represents a duration, the difference between two dates or times.
 |  
 |  Timedelta is the pandas equivalent of python's ``datetime.timedelta``
 |  and is interchangeable with it in most cases.
 |  
 |  Parameters
 |  ----------
 |  value : Timedelta, timedelta, np.timedelta64, string, or integer
 |  unit : str, default 'ns'
 |      Denote the unit of the input, if input is an integer.
 |  
 |      Possible values:
 |  
 |      * 'Y', 'M', 'W', 'D', 'T', 'S', 'L', 'U', or 'N'
 |      * 'days' or 'day'
 |      * 'hours', 'hour', 'hr', or 'h'
 |      * 'minutes', 'minute', 'min', or 'm'
 |      * 'seconds', 'second', or 'sec'
 |      * 'milliseconds', 'millisecond', 'millis', or 'milli'
 |      * 'microseconds', 'microsecond', 'micros', or 'micro'
 |      * 'nanoseconds', 'nanosecond', 'nanos', 'nano', or 'ns'.

a+pd.Timedelta(2,unit='days')

Timestamp('2020-07-23 12:12:01')

a+pd.Timedelta(2,unit='m')

Timestamp('2020-07-21 12:14:01')

pd.DateOffset

help(pd.DateOffset)`也具有的類似的功能：但是需要注意關(guān)鍵詞單復(fù)數(shù)的區(qū)別：

Parameters that **add** to the offset (like Timedelta):
 |  
 |      - years
 |      - months
 |      - weeks
 |      - days
 |      - hours
 |      - minutes
 |      - seconds
 |      - microseconds
 |      - nanoseconds
 |  
Parameters that **replace** the offset value:
 |  
 |      - year
 |      - month
 |      - day
 |      - weekday
 |      - hour
 |      - minute
 |      - second
 |      - microsecond
 |      - nanosecond.
    

a+pd.DateOffset(days=1)

Timestamp('2020-07-22 12:12:01')

b+pd.DateOffset(day=1)

Timestamp('2020-07-01 12:12:23')