熊猫冷凝分组

Question

我目前正在尝试使用Pandas中的groupby函数来聚合一些CSV数据。

这是我目前在CSV中的一个小样本：

Company,School,Number,Type
Adtelem Global Education Inc.,Carrington,3,For-Profit
Adtelem Global Education Inc.,Carrington,4,For-Profit
Adtelem Global Education Inc.,Carrington,1,For-Profit
Adtelem Global Education Inc.,Carrington,4,For-Profit
Adtelem Global Education Inc.,Carrington,3,For-Profit
Adtelem Global Education Inc.,Carrington,3,For-Profit
Adtelem Global Education Inc.,DeVry Institute of Technology,4,For-Profit
Adtelem Global Education Inc.,DeVry Institute of Technology,4,For-Profit
Adtelem Global Education Inc.,DeVry Institute of Learning,16,   For-Profit
Adtelem Global Education Inc.,DeVry Institute of Learning,9,    
Career Education Corporation,Le Cordon Blue College of Culinary Arts,6,For-Profit
Career Education Corporation,Le Cordon Blue College of Culinary Arts,23,For-Profit

目前，有许多重复的“学校”栏目(卡林顿，德弗里等)，我想把它们压缩下来。更具体地说，我希望每一所独特的学校都有1行，其中也包括该学校所有实例的数字，但保留拥有该学校的公司的名称(第一栏)和学校的类型(最后一栏)。

最终产品如下所示：

Company,School,Number,Type
Adtelem Global Education Inc.,Carrington,18,For-Profit,
Adtelem Global Education Inc., DeVry Institute of Technology,8,For-Profit
Adtelem Global Education Inc.,DeVry Institute of Learning,25,For-Profit
Career Education Corporation,Le Cordon Blue College of Culinary Arts,29,For-Profit

我使用了以下代码：

data2 = data.groupby("School").sum()

然而，当我这样做，我也失去了公司和类型附属于每一所学校。我知道解决方案是相当基本的，但我是新的潘达斯，所以任何帮助你们都可以提供将非常感谢！

Answer 1

我会用groupby + agg来做这个

df.groupby('School', as_index=False)\
    .agg({'Company' : 'first', 'Type' : 'first', 'Number' : 'sum'})

                                    School                        Company  \
0                               Carrington  Adtelem Global Education Inc.   
1              DeVry Institute of Learning  Adtelem Global Education Inc.   
2            DeVry Institute of Technology  Adtelem Global Education Inc.   
3  Le Cordon Blue College of Culinary Arts   Career Education Corporation   

   Number        Type  
0      18  For-Profit  
1      25  For-Profit  
2       8  For-Profit  
3      29  For-Profit 

我认为最好显式地聚合所有列。

Answer 2

您可以提供要按

data2 = data.groupby(["School", "Company", "Type"]).sum()