代码之家 › 专栏 › 技术社区 › moshevi

大熊猫按任意时间段统计剩余时间

pandas python

moshevi · 技术社区 · 8 年前

我想把我的 dataframe 到任意时间段,然后创建列,计算每行的剩余时间,直到下一个n个时间段结束。

例如,此输入:

df = pd.DataFrame({'dates':pd.date_range('2017-01-01',periods=8,freq='1d')}).set_index('dates')

# the inclusive ends of the time periods
rolling_dates = ['2017-01-02', '2017-01-05', '2017-01-07', '2017-01-08']  

periods_offests = [0, 1, 2]  # the remaining time periods columns

将产生以下输出:

dates         periods_expiry_days_0 periods_expiry_days_1   periods_expiry_days_2
2017-01-01            1                     4.0                       6.0
2017-01-02            0                     3.0                       5.0
2017-01-03            2                     4.0                       5.0
2017-01-04            1                     3.0                       4.0
2017-01-05            0                     2.0                       3.0
2017-01-06            1                     2.0                       Nan
2017-01-07            0                     1.0                       Nan
2017-01-08            0                     Nan                       Nan

1 回复 | 直到 8 年前

T. Ray 8 年前

def map_offsets(x):
    '''Calculate the day offsets'''
    days = [x for x in (pd_rolling - x).days if x >= 0]
    days += [np.nan] * (len(pd_rolling) - len(days))

    return [days[i] for i in periods_offsets]

df = pd.DataFrame({'dates': pd.date_range('2017-01-01', periods=8, freq='1d')}).set_index('dates')

# the inclusive ends of the time periods
rolling_dates = ['2017-01-02', '2017-01-05', '2017-01-07', '2017-01-08']

# the remaining time periods columns
periods_offsets = [0, 1, 2]

# Added: Casting to datetime will make offset calculation easier
pd_rolling = pd.to_datetime(rolling_dates)

# Create new column names for periods
fmt = 'periods_expiry_days_{}'
columns = [fmt.format(x) for x in periods_offsets]

# Subtract index from rolling date values, and add to dataframe
df[columns] = pd.DataFrame(df.index.map(map_offsets).tolist(), index=df.index)

推荐文章