Python 根据日期时间条件乘以数据帧行

我有以下示例数据帧：

df = pd.DataFrame(columns=['Total_concentration'], index=pd.to_datetime([]))
df.loc[pd.Timestamp(year=2017, month=1, day=1, hour=12)] = pd.Series([1], ['Total_concentration'])
df.loc[pd.Timestamp(year=2017, month=1, day=2, hour=12)] = pd.Series([2], ['Total_concentration'])
df.loc[pd.Timestamp(year=2017, month=1, day=3, hour=12)] = pd.Series([3], ['Total_concentration'])
df.loc[pd.Timestamp(year=2017, month=1, day=4, hour=12)] = pd.Series([1], ['Total_concentration'])
df.loc[pd.Timestamp(year=2017, month=1, day=5, hour=12)] = pd.Series([2], ['Total_concentration'])
df.loc[pd.Timestamp(year=2017, month=1, day=6, hour=12)] = pd.Series([3], ['Total_concentration'])

我想添加一列“总通量”，即2017-01-02和2017-01-03日期之间的“总浓度”列乘以2，以及2017-01-04和2017-01-05日期之间的“总浓度”列乘以3。即，生成的数据帧如下所示：

df2 = pd.DataFrame(columns=['Total_concentration','Total_flux'], index=pd.to_datetime([]))
df2.loc[pd.Timestamp(year=2017, month=1, day=1, hour=12)] = pd.Series([1,1], ['Total_concentration','Total_flux'])
df2.loc[pd.Timestamp(year=2017, month=1, day=2, hour=12)] = pd.Series([2,4], ['Total_concentration','Total_flux'])
df2.loc[pd.Timestamp(year=2017, month=1, day=3, hour=12)] = pd.Series([3,6], ['Total_concentration','Total_flux'])
df2.loc[pd.Timestamp(year=2017, month=1, day=4, hour=12)] = pd.Series([1,3], ['Total_concentration','Total_flux'])
df2.loc[pd.Timestamp(year=2017, month=1, day=5, hour=12)] = pd.Series([2,6], ['Total_concentration','Total_flux'])
df2.loc[pd.Timestamp(year=2017, month=1, day=6, hour=12)] = pd.Series([3,3], ['Total_concentration','Total_flux'])

---

执行此操作最有效的方法是什么？

使用切片

DataetimeIndex

的解决方案，然后用原始列替换缺少的值：

df.loc['2017-01-02':'2017-01-03', 'Total_flux'] = df.loc['2017-01-01':'2017-01-03', 'Total_concentration'] * 2
df.loc['2017-01-04':'2017-01-05', 'Total_flux'] = df.loc['2017-01-04':'2017-01-05', 'Total_concentration'] * 3

df['Total_flux'] = df['Total_flux'].fillna(df['Total_concentration'])
print (df)
                    Total_concentration  Total_flux
2017-01-01 12:00:00                   1           1
2017-01-02 12:00:00                   2           4
2017-01-03 12:00:00                   3           6
2017-01-04 12:00:00                   1           3
2017-01-05 12:00:00                   2           6
2017-01-06 12:00:00                   3           3

或可能用于多个和的值：

---

使用以下代码：

df['Total_flux'] = df.Total_concentration.multiply(pd.cut(df.index.to_series(),
    [pd.Timestamp('2017-01-02'), pd.Timestamp('2017-01-04'), pd.Timestamp('2017-01-06')],
    right=False, labels=[2, 3]).dropna().astype(int), fill_value=1)

计算新列的想法是：

• 将df的索引剪切到多个箱子中：
  • 2017-01-02和2017-01-04之间，标签为2
  • 2017-01-04和2017-01-06之间，标签为3 （两个范围都在右边缘打开）
• 将（分类）结果转换为int
• 用上述结果乘以总_浓度 填充值为1

---

以下方法可以将子集作为离散序列，但我不确定如何将其放入数据框架：总通量=np.乘法（总浓度['2017-01-02'：'2017-01-03']，2）

df['Total_flux'] = df.Total_concentration.multiply(pd.cut(df.index.to_series(),
    [pd.Timestamp('2017-01-02'), pd.Timestamp('2017-01-04'), pd.Timestamp('2017-01-06')],
    right=False, labels=[2, 3]).dropna().astype(int), fill_value=1)