如何聚合 CSV 中的行，排除基于值列表的行答案

【问题标题】：How to aggregate rows from a CSV, excluding ones based on a list of values如何聚合 CSV 中的行，排除基于值列表的行
【发布时间】：2022-11-17 14:20:56
【问题描述】：

我有一个包含以下数据的 csv：

"id","Title","Author(s)","Format","Size","Tags"
"1","Horse","John","KFX","122","Classic"
"1","Horse","John","KFX","122","Drama"
"1","Horse","John","KFX","122","Horror"
"1","Horse","John","AZW3","122","Classic"
"1","Horse","John","AZW3","122","Drama"
"1","Horse","John","AZW3","122","Horror"
"1","Horse","John","PDF","122","Classic"
"1","Horse","John","PDF","122","Drama"
"1","Horse","John","PDF","122","Horror"
"2","Banana","Anna","AZW3","312","SciFi"
"2","Banana","Julia","AZW3","312","SciFi"
"2","Banana","Anna","PDF","312","SciFi"
"2","Banana","Julia","PDF","312","SciFi"

我想，使用熊猫，得到这个：

"id","Title","Author(s)","Format","Size","Tags"
"1","Horse","John","KFX","122","Classic, Drama, Horror"
"2","Banana","Anna, Julia","AZW3","312","SciFi"

它将根据列表的值排除行的位置 KFX, AZW3, PDF

如果它有 KFX，则排除其他两个。如果没有 KFX，请检查列表中的下一个 AZW3，如果存在，则排除 PDF（如果存在）

然后，汇总作者和标签。

我正在寻找这样做的方法，但到目前为止没有成功。我试图按 id 分组，然后根据列表（KFX、AZW3、PDF）排除行，然后聚合。

我能够使用以下方法进行聚合：

df.groupby(['id']).agg(lambda x: ','.join(x))

但是有两个问题：首先，它将包括包含所有“格式”的行。其次，它也会重复其他列 “1、1、1”、“马、马、马”、“约翰、约翰、约翰”

等等。

【问题讨论】：

标签： python pandas

【解决方案1】：

您可以使用自定义函数来聚合：

def uniq_str(l):
    return ', '.join(dict.fromkeys(map(str, l)))

def agg_format(l):
    s = set(l) # not necessary if only 1 comparison
    if 'KFX' in s:
        return 'KFX'
    else:
        exclude = {'PDF'}
        return next((x for x in l if x not in exclude), None)
    
df = pd.read_csv('file.csv')
    
agg = {c: uniq_str for c in df}
agg['Format'] = agg_format

import csv 
(df.groupby('id', as_index=False).agg(agg)
   .to_csv('out.csv', index=False, quoting=csv.QUOTE_ALL)
)

输出：

"id","Title","Author(s)","Format","Size","Tags"
"1","Horse","John","KFX","122","Classic, Drama, Horror"
"2","Banana","Anna, Julia","AZW3","312","SciFi"

【讨论】：

惊人！！有用。太感谢了！我没有足够的编程经验，我花了一整天的时间来解决这个问题。

【解决方案2】：

def function1(dd:pd.DataFrame):
    lambda1=lambda x:dd.query("Format==@x").groupby('id,Title,Format,Size'.split(','),as_index=False).agg(set)
    
    if len(dd.query("Format=='KFX'"))>0:
        dd=lambda1('KFX')
    elif len(dd.query("Format=='AZW3'"))>0:
        dd=lambda1('AZW3')
    else:
        dd=lambda1('PDF')
    return dd.applymap(lambda x:','.join(x) if isinstance(x,set) else x)
df1.groupby('id').apply(function1).set_index('id').rename_axis(None)

  Title Format  Size   Author(s)                  Tags
1   Horse    KFX   122        John  Drama,Classic,Horror
2  Banana   AZW3   312  Anna,Julia                 SciFi

【讨论】：