如果有條件，如何在 Pandas 的新列中重復值

Question

我有一個這樣的 dataframe：

Index     Category   Sales   
01-01-21    Blue     83.20  
01-02-21    Yellow   98.57  
01-03-21    Green    55.70  
01-04-21    Orange   87.32  
01-05-21    Black    65.02  
01-06-21    Yellow   61.05  
01-07-21    White    10.54  
01-08-21    Yellow   73.59  
01-09-21    Blue     13.88  
01-10-21    Black    47.39  
01-11-21    Blue     55.75  
01-12-21    Yellow   39.64  
01-01-22    Green    39.64  
01-02-22    Orange   39.64  
01-03-22    Black    28.87    
01-04-22    Yellow   82.40   
01-05-22    White    28.09    
01-06-22    Yellow   74.64  
01-07-22    Blue     18.27   
01-08-22    Black    11.16    
01-09-22    Green    79.49  
01-10-22    Green    46.74    
01-11-22    Yellow   6.81    
01-12-22    Blue     8.00    

我想添加一個新列，在接下來的 12 個月中重復 10 月、11 月和 12 月的銷售額值。 所以結果看起來像這樣：

Index   Category     Sales   New Column 
01-01-21    Blue     83.20  
01-02-21    Yellow   98.57  
01-03-21    Green    55.70  
01-04-21    Orange   87.32  
01-05-21    Black    65.02  
01-06-21    Yellow   61.05  
01-07-21    White    10.54  
01-08-21    Yellow   73.59  
01-09-21    Blue     13.88  
+--------------------------------------------+
01-10-21    Black    47.39  
01-11-21    Blue     55.75  
01-12-21    Yellow   39.64  
+--------------------------------------------+
01-01-22    Green    39.64   47.39 
01-02-22    Orange   39.64   55.75 
01-03-22    Black    28.87   39.64 
01-04-22    Yellow   82.40   47.39 
01-05-22    White    28.09   55.75 
01-06-22    Yellow   74.64   39.64 
01-07-22    Blue     18.27   47.39 
01-08-22    Black    11.16   55.75 
01-09-22    Green    79.49   39.64 
01-10-22    Green    46.74   47.39 
01-11-22    Yellow   6.81    55.75 
01-12-22    Blue     8.00    39.64 

我試過這種方式：

recurr_oct = new.loc[new.index.isin(['01-10-21']), 'Sales'].values
recurr_nov = new.loc[new.index.isin(['01-11-21']), 'Sales'].values
recurr_dec = new.loc[new.index.isin(['01-12-21']), 'Sales'].values

new.index = pd.to_datetime(new.index)
rating = []
for row in new.index.month:
    if row == 1 or row == 4 or row == 7 or row == 10:    rating.append(recurr_oct)
    elif row == 2 or row == 5 or row == 8 or row == 11:   rating.append(recurr_nov)
    elif row == 3 or row == 6 or row == 9 or row == 12:  rating.append(recurr_dec)
    else: rating.append(0)
        
new['New Column'] = rating
new.head()

但這需要時間，而且看起來不太好，有沒有更好的方法呢？

謝謝！

Answer 1

為此，我建議將模數%與.month屬性結合使用。

要更改您的代碼，我會使用這樣的東西：

# your code...
for row in new.index.month:
    if (row % 3) == 1:    rating.append(recurr_oct)
    elif (row % 3) == 2:   rating.append(recurr_nov)
    elif (row % 3) == 0:  rating.append(recurr_dec)

您也可以將它變成 function，然后將df.apply()與該 function 一起使用。恕我直言，更清潔，可能更快。

Answer 2

if row % 3 == 1: rating.append(recurr_oct)
elif row % 3 == 2: rating.append(nov)
else: rating.append(recurr_dec)