datetime64[ns] 的逻辑语句

Question

我在使用df.loc 时遇到以下错误。我有一个参数 s，它是 datetime64[ns]，我有一个数据框，其中有两个列 MDT1 和 MTD2，它也是 datetime64[ns] 类型。当我使用 df.loc 使用以下语句时

s= df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
s = pd.to_datetime(s)
df_1x2.loc[(s>df_1x2['MDT1'])&(s<df_1x2['MDT2']),'Morning']= s

代码的最后一行显示 Jupyter Notebook 中的以下错误

ValueError: Can only compare identically-labeled Series objects

但是，当我第二次运行时，它按预期工作。只是在 Jupyter Notebook 中的第一次运行让我遇到了错误。

以下是示例数据文件的下载链接：

Click here to download the sample data file

下面是代码：

s = pd.to_datetime(df['Date']).astype(str)+' '+df['Time.1'].astype(str)+':00'
df['Matchdate'] = pd.to_datetime (s)

d =np.reciprocal(df_1x2[['Home','Draw','Away']].astype(float))

df_1x2['Margin'] = d.sum(axis = 1)-1
df_1x2['Open_DT'] = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
df_1x2['Open_DT'] = pd.to_datetime(df_1x2['Open_DT'])
df_1x2['Open_DT1'] = df_1x2.groupby(['Match','Odds_Type'])['Open_DT'].transform(min)
s = df_1x2['Open_DT'] == df_1x2['Open_DT1']
df_1x2.loc[s,'Open'] ='Opening'


df_1x2['Current_DT'] = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
df_1x2['Current_DT'] = pd.to_datetime(df_1x2['Current_DT'])
df_1x2['Current_DT1'] = df_1x2.groupby(['Match','Odds_Type'])['Current_DT'].transform(max)
sc = df_1x2['Current_DT'] == df_1x2['Current_DT1']
df_1x2.loc[sc,'Current'] ='Current'
df_1x2.loc[df_1x2.Open == 'Opening','Current']='Opening'

df_1x2 ['morning Date'] = pd.Timestamp.today()
Time1 = '08:00:00'
Time2 = '11:00:00'
s = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
s = pd.to_datetime(s)
df_1x2.reset_index(drop = True, inplace = True)

df_1x2 ['morning Date'] = pd.to_datetime(df_1x2 ['morning Date']).dt.date
df_1x2['MDT1'] = pd.to_datetime(df_1x2['morning Date'].astype(str)+' '+Time1)


df_1x2['MDT2'] = pd.to_datetime(df_1x2['morning Date'].astype(str)+' '+Time2)

df_1x2 .reset_index(drop= True, inplace = True)



df_1x2.loc[(s>df_1x2['MDT1'])&(s<df_1x2['MDT2']),'Morning']= s

我该如何纠正它？

Answer 1

您正在将整个系列 s 应用于过滤的数据框列。我认为使用mask 或np.where 而不是.loc 更容易：

df_1x2['Morning'] = df_1x2['Morning'].mask((s > df_1x2['MDT1']) & (s < df_1x2['MDT2']), s)

或np.where:

df_1x2['Morning'] = np.where((s > df_1x2['MDT1']) & (s < df_1x2['MDT2']), s, df_1x2['Morning'])

---

如果该列尚不存在，则使用：

df_1x2['Morning'] = np.where((s > df_1x2['MDT1']) & (s < df_1x2['MDT2']), s, np.datetime64('NaT'))

如果您分配像10 这样的单个值而不是长度比过滤数据帧长的系列，则.loc 可以用于此用例。

---

问题是您在定义s 之后reset_index()，这意味着它与您的数据框无法比较：

s = pd.to_datetime(df['Date']).astype(str)+' '+df['Time.1'].astype(str)+':00'
df['Matchdate'] = pd.to_datetime (s)

d =np.reciprocal(df_1x2[['Home','Draw','Away']].astype(float))

df_1x2['Margin'] = d.sum(axis = 1)-1
df_1x2['Open_DT'] = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
df_1x2['Open_DT'] = pd.to_datetime(df_1x2['Open_DT'])
df_1x2['Open_DT1'] = df_1x2.groupby(['Match','Odds_Type'])['Open_DT'].transform(min)
s = df_1x2['Open_DT'] == df_1x2['Open_DT1']
df_1x2.loc[s,'Open'] ='Opening'


df_1x2['Current_DT'] = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
df_1x2['Current_DT'] = pd.to_datetime(df_1x2['Current_DT'])
df_1x2['Current_DT1'] = df_1x2.groupby(['Match','Odds_Type'])['Current_DT'].transform(max)
sc = df_1x2['Current_DT'] == df_1x2['Current_DT1']
df_1x2.loc[sc,'Current'] ='Current'
df_1x2.loc[df_1x2.Open == 'Opening','Current']='Opening'

df_1x2 ['morning Date'] = pd.Timestamp.today()
Time1 = '08:00:00'
Time2 = '11:00:00'

########### I moved s from here...........

df_1x2.reset_index(drop = True, inplace = True)

df_1x2 ['morning Date'] = pd.to_datetime(df_1x2 ['morning Date']).dt.date
df_1x2['MDT1'] = pd.to_datetime(df_1x2['morning Date'].astype(str)+' '+Time1)


df_1x2['MDT2'] = pd.to_datetime(df_1x2['morning Date'].astype(str)+' '+Time2)

df_1x2 .reset_index(drop= True, inplace = True)

########### .........to here. 
##### Resetting the  index after defining s, means you can no longer compare directly to your dataframe as s has a different index than your dataframe now.

s = df_1x2['date'].astype (str)+'-'+ df_1x2['Time'].astype(str)
s = pd.to_datetime(s)


df_1x2.loc[(s>df_1x2['MDT1'])&(s<df_1x2['MDT2']),'Morning']= s