[Pandas Day9]时序数据

ex 9-1

import pandas as pd
import numpy as np
df = pd.read_csv('data/time_series_one.csv', parse_dates=['日期'])
# print(df.head())

# 销售额出现最大值的是星期几？（提示：利用dayofweek函数）
print(df['日期'].dt.dayofweek[df['销售额'].idxmax()])

# 计算除去春节、国庆、五一节假日的月度销售总额
holiday = pd.date_range(start='20170501', end='20170503').append(
          pd.date_range(start='20171001', end='20171007')).append(
          pd.date_range(start='20180215', end='20180221')).append(
          pd.date_range(start='20180501', end='20180503')).append(
          pd.date_range(start='20181001', end='20181007')).append(
          pd.date_range(start='20190204', end='20190224')).append(
          pd.date_range(start='20190501', end='20190503')).append(
          pd.date_range(start='20191001', end='20191007'))
result = df[~df['日期'].isin(holiday)].set_index('日期').resample('MS').sum()
print(result.head())

# 按季度计算周末（周六和周日）的销量总额
result = df[df['日期'].dt.dayofweek.isin([5,6])].set_index('日期').resample('QS').sum()
print(result.head())

# 从最后一天开始算起，跳过周六和周一，以5天为一个时间单位向前计算销售总和
df_temp = df[~df['日期'].dt.dayofweek.isin([5,6])].set_index('日期').iloc[::-1]
L_temp,date_temp = [],[0]*df_temp.shape[0]
for i in range(df_temp.shape[0]//5):
    L_temp.extend([i]*5)
L_temp.extend([df_temp.shape[0]//5]*(df_temp.shape[0]-df_temp.shape[0]//5*5))
date_temp = pd.Series([i%5==0 for i in range(df_temp.shape[0])])
df_temp['num'] = L_temp
result = pd.DataFrame({'5天总额':df_temp.groupby('num')['销售额'].sum().values},
                       index=df_temp.reset_index()[date_temp]['日期']).iloc[::-1]
print(result.head())

# 假设现在发现数据有误，所有同一周里的周一与周五的销售额记录颠倒了，请计算2018年中每月第一个周一的销售额（如果该周没有周一或周五的记录就保持不动）
from datetime import datetime
df_temp = df.copy()
df_fri = df.shift(4)[df.shift(4)['日期'].dt.dayofweek==1]['销售额']
df_mon = df.shift(-4)[df.shift(-4)['日期'].dt.dayofweek==5]['销售额']
df_temp.loc[df_fri.index,['销售额']] = df_fri
df_temp.loc[df_mon.index,['销售额']] = df_mon
print(df_temp.loc[df_temp[df_temp['日期'].dt.year==2018]['日期'][
        df_temp[df_temp['日期'].dt.year==2018]['日期'].apply(
        lambda x:True if datetime.strptime(str(x).split()[0],'%Y-%m-%d').weekday() == 0
        and 1 <= datetime.strptime(str(x).split()[0],'%Y-%m-%d').day <= 7 else False)].index,:])

ex9-2

import pandas as pd
# 以50天为窗口计算滑窗均值和滑窗最大值（min_periods设为1）
df = pd.read_csv('data/time_series_one.csv',index_col='日期',parse_dates=['日期'])
print(df['销售额'].rolling(window=50,min_periods=1).mean().head())
print(df['销售额'].rolling(window=50,min_periods=1).max().head())

# 现在有如下规则：若当天销售额超过向前5天的均值，则记为1，否则记为0，请给出2018年相应的计算结果
def f(x):
    if len(x) == 6:
        return 1 if x[-1]>np.mean(x[:-1]) else 0
    else:
        return 0
result_b = df.loc[pd.date_range(start='20171227',end='20181231'),:].rolling(
                                                    window=6,min_periods=1).agg(f)[5:].head()
print(result_b.head())

# 将(c)中的“向前5天”改为“向前非周末5天”，请再次计算结果
def f(x):
    if len(x) == 8:
        return 1 if x[-1]>np.mean(x[:-1][pd.Series([
            False if i in [5,6] else True for i in x[:-1].index.dayofweek],index=x[:-1].index)]) else 0
    else:
        return 0
result_c = df.loc[pd.date_range(start='20171225',end='20181231'),:].rolling(
                                    window=8,min_periods=1).agg(f)[7:].head()
print(result_c.head())