熊猫选择匹配多列

Question

我有这样的数据：

category = ['Car','Car','Car','Car','Truck','Truck','Truck']
name = ['Camry','Camry','Camry','Camry','Tacoma','Tundra','Tundra']
year = ['2007','2007','2008','2009','2010','2010','2011']
vals = [0.1,0.5,0.2,0.9,0.8,0.4,0.9]
df = pd.DataFrame({'Category': category,
                   'Name': name,
                   'Year': year,
                   'Vals': vals})

index	Category	Name	Year	Vals
0	Car	Camry	2007	0.1
1	Car	Camry	2007	0.5
2	Car	Camry	2008	0.2
3	Car	Camry	2009	0.9
4	Truck	Tacoma	2010	0.8
5	Truck	Tundra	2010	0.4
6	Truck	Tundra	2011	0.9

然后，我有一组（类别、名称、年份）组合，我想为其过滤数据框。它们可以是任何格式，但在这里它们位于数据框中。

combinations_i_want = pd.DataFrame()
# (Car, Camry, 2007)
combinations_i_want = combinations_i_want.append({'Category':'Car', 'Name':'Camry','Year':'2007'},ignore_index=True) # 2 matches in df
# (Truck, Tundra, 2010)
combinations_i_want = combinations_i_want.append({'Category':'Truck', 'Name':'Tundra','Year':'2010'},ignore_index=True) # 1 match in df

我想提取 df 中与这两种组合完全匹配的行。这些将是第 0、1 和 5 行。结果表如下所示：

index	Category	Name	Year	Vals
0	Car	Camry	2007	0.1
1	Car	Camry	2007	0.5
5	Truck	Tundra	2010	0.4

注意：我不需要旧索引，它们只是为了帮助可视化。

我该怎么做？

Answer 1

使用数据框查询，它将根据布尔逻辑为您提供完美匹配

print(df.query("(Category=='Car' and Name=='Camry' and Year=='2007') or (Category=='Truck' and Name=='Tundra' and Year=='2010')"))

输出：

     Category    Name  Year  Vals
   0      Car   Camry  2007   0.1
   1      Car   Camry  2007   0.5
   5    Truck  Tundra  2010   0.4

Answer 2

您可以简单地右加入您想要的列。

result = df.merge(combinations_i_want, how='right', on=['Category', 'Name', 'Year'])

Answer 3

您应该使用.loc 和.isin 而不是.append

你的句子可能是这样的：

df.loc[(df['Category'].isin(['Car', 'Truck'])) & (df['Name'].isin(['Camry', 'Tundra'])) & (df['Year'].isin(['2007', '2010']))]

这应该会产生您期望的结果。

如果需要，您可以将其分配给变量，例如

combinations_i_want = df.loc[(df['Category'].isin(['Car', 'Truck'])) &
         (df['Name'].isin(['Camry', 'Tundra'])) &
         (df['Year'].isin(['2007', '2010']))]
    
print(combinations_i_want)