Python pandas，分组 groupby()，分组后的聚合函数，交叉表 crosstab()

约定不等于承诺〃 2022-04-23 12:42 184阅读 0赞

**demo.py（分组，groupby()，分组后的聚合函数）：**

import pandas as pd
    
    
    my_list = [{"name":"zhangsan","age":18, "province":"jiangsu"}, {"name":"lisi", "age":19, "province":"henan"}, {"name":"xiaohong", "age":18, "province":"henan"}, {"name":"wangwu", "age":18, "province":"jiangsu"}]
    df = pd.DataFrame(my_list)
    print(df)
    '''
       age      name province
    0   18  zhangsan  jiangsu
    1   19      lisi    henan
    2   18  xiaohong    henan
    3   18    wangwu  jiangsu
    '''
    
    
    # 分组
    grouped = df.groupby(by="province")  # 根据"province"列分组
    print(type(grouped))  # <class 'pandas.core.groupby.generic.DataFrameGroupBy'>  可以遍历
    
    for i,j in grouped:
        print(i)  # henan    jiangsu
        print(j)  # DataFrame类型
        print("-"*30)
    '''
    henan
       age      name province
    1   19      lisi    henan
    2   18  xiaohong    henan
    ------------------------------
    jiangsu
       age      name province
    0   18  zhangsan  jiangsu
    3   18    wangwu  jiangsu
    ------------------------------
    '''
    
    # 也可以通过bool索引达到分组的效果。
    print(df[df["province"]=="henan"])
    '''
       age      name province
    1   19      lisi    henan
    2   18  xiaohong    henan
    '''
    
    # 聚合函数。 count()每一列的个数
    print(grouped.count())  # DataFrame类型
    '''
              age  name
    province
    henan       2     2
    jiangsu     2     2
    '''
    
    # 统计分组后指定列的个数
    print(grouped["name"].count())  # Series类型
    '''
    province
    henan      2
    jiangsu    2
    Name: name, dtype: int64
    '''

**DataFrameGroupBy分组对象的聚合函数：**

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2hvdXlhbmh1YTE_size_16_color_FFFFFF_t_70][]

**demo.py（按多列进行分组，groupby()）：**

import pandas as pd
    
    
    my_list = [{"name":"zhangsan","age":18, "province":"jiangsu"}, {"name":"lisi", "age":19, "province":"henan"}, {"name":"xiaohong", "age":18, "province":"henan"}, {"name":"wangwu", "age":18, "province":"jiangsu"}]
    df = pd.DataFrame(my_list)
    print(df)
    '''
       age      name province
    0   18  zhangsan  jiangsu
    1   19      lisi    henan
    2   18  xiaohong    henan
    3   18    wangwu  jiangsu
    '''
    
    
    # 根据多列分组
    grouped = df.groupby(by=["province","age"])  # 根据"province"和"age"列分组
    
    for i,j in grouped:
        print(i)  # ('henan', 18)  ('henan', 19)  ('jiangsu', 18)
        print(j)  # DataFrame类型
        print("-"*30)
    '''
    ('henan', 18)
       age      name province
    2   18  xiaohong    henan
    ------------------------------
    ('henan', 19)
       age  name province
    1   19  lisi    henan
    ------------------------------
    ('jiangsu', 18)
       age      name province
    0   18  zhangsan  jiangsu
    3   18    wangwu  jiangsu
    ------------------------------
    '''
    
    
    # 根据多列分组
    # grouped = df["name"].groupby(by=["province","age"])  # 错误。 df["name"]是Series类型，没有"province"和"age"列。
    grouped2 = df["name"].groupby(by=[df["province"],df["age"]])   # 正确。 by=[df["province"],df["age"]表示df的"province"列和df的"age"列
    print(grouped2.count())  # Series类型。 ("province"和"age"都是索引(复合索引)，真正的数据只有一列)
    '''
    province  age
    henan     18     1
              19     1
    jiangsu   18     2
    Name: name, dtype: int64
    '''
    # 如果想得到DataFrame类型可以使用 df[["name"]]代替df["name"]
    
    
    grouped3 = df.groupby(by=[df["province"],df["age"]])["name"]  # 取"name"列也可以在分组之后取。
    print(grouped3.count())
    '''
    province  age
    henan     18     1
              19     1
    jiangsu   18     2
    Name: name, dtype: int64
    '''

**demo.py（交叉表，特殊的分组工具，crosstab()）：**

import pandas as pd
    
    # 模拟用户购买商品的表(数据)
    my_list = [{"user_id": 11, "goods": "苹果"}, {"user_id": 11, "goods": "苹果"}, {"user_id": 11, "goods": "香蕉"},
               {"user_id": 22, "goods": "苹果"}, {"user_id": 22, "goods": "香蕉"}, {"user_id": 22, "goods": "香蕉"},
               {"user_id": 33, "goods": "梨"}, {"user_id": 33, "goods": "香蕉"}, {"user_id": 33, "goods": "苹果"}]
    df = pd.DataFrame(my_list)
    print(df)
    '''
      goods     user_id
    0    苹果       11
    1    苹果       11
    2    香蕉       11
    3    苹果       22
    4    香蕉       22
    5    香蕉       22
    6     梨        33
    7    香蕉       33
    8    苹果       33
    '''
    
    # 交叉表 (特殊的分组工具)
    crass_tb = pd.crosstab(df["user_id"], df["goods"])  # 交叉表，统计每个用户购买各种商品的数量
    print(crass_tb)
    '''
    goods    梨  苹果  香蕉
    user_id           
    11       0    2    1
    22       0    1    2
    33       1    1    1
    '''

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2hvdXlhbmh1YTE_size_16_color_FFFFFF_t_70]: /images/20220318/9d8288fce2a142f9b702940d967db8d8.png