logo标签列表

如何根据层次结构计算列值

pythonpandasnumpy
6
假设我们同意以下层级顺序。
婴儿 -> 儿童 -> 青少年 -> 成人
我有这组数据集。
           Name         Stage  Highest_Stage_Reached
0          Adam         Child  
1         Barry         Child
2           Ben         Adult
3          Adam      Teenager
4         Barry         Adult
5           Ben         Baby

我该如何设置数据集来填充“Highest_Stage_Reached”字段呢？
           Name         Stage  Highest_Stage_Reached
0          Adam         Child  Teenager
1         Barry         Child  Adult
2           Ben         Adult  Adult
3          Adam      Teenager  Teenager
4         Barry         Adult  Adult
5           Ben         Baby   Adult
3个回答
3

您可以使用：

d={'Baby':0,'Child':1,'Teenager':2,'Adult':3}
df['rank']=df.Stage.map(d)
df['Highest_Stage_Reached']=df.groupby('Name')['rank'].transform('max').\
                                         map({v: k for k, v in d.items()})
print(df.drop('rank',1))

    Name     Stage Highest_Stage_Reached
0   Adam     Child              Teenager
1  Barry     Child                 Adult
2    Ben     Adult                 Adult
3   Adam  Teenager              Teenager
4  Barry     Adult                 Adult
5    Ben      Baby                 Adult
1
将您的层次结构放入列表中，使用列表的索引。
l = ['Baby', 'Child', 'Teenager', 'Adult']
df = pd.DataFrame({'Name': ['Adam', 'Barry', 'Ben', 'Adam', 'Barry', 'Ben'], 'Stage': ['Child', 'Child', 'Adult', 'Teenager', 'Adult', 'Baby']})

cond = [df['Stage'] == 'Baby',df['Stage'] == 'Child',df['Stage'] == 'Teenager',df['Stage'] == 'Adult']
df['Highest_Stage_Reached'] = np.select(cond, [0,1,2,3])

    Name     Stage  Highest_Stage_Reached
0   Adam     Child                      1
1  Barry     Child                      1
2    Ben     Adult                      3
3   Adam  Teenager                      2
4  Barry     Adult                      3
5    Ben      Baby                      0

df['Highest_Stage_Reached'] = (df.groupby('Name')['Highest_Stage_Reached'].transform(max))

    Name     Stage  Highest_Stage_Reached
0   Adam     Child                      2
1  Barry     Child                      3
2    Ben     Adult                      3
3   Adam  Teenager                      2
4  Barry     Adult                      3
5    Ben      Baby                      3


df['Highest_Stage_Reached'] = df['Highest_Stage_Reached'].apply(lambda x: l[x])
print(df)

输出：

    Name     Stage Highest_Stage_Reached
0   Adam     Child              Teenager
1  Barry     Child                 Adult
2    Ben     Adult                 Adult
3   Adam  Teenager              Teenager
4  Barry     Adult                 Adult
5    Ben      Baby                 Adult
1
将该列转换为分类变量，使用有序参数。现在可以进行排序。这也将支持阶段中可变数量的参数。
df['Stage'] = pd.Categorical(df['Stage'], ordered=True, categories=['Baby', 'Child','Teenager','Adult'])

df['Highest_Stage_Reached'] = df.groupby('Name').Stage.transform('max')

    Name    Stage       Highest_Stage_Reached
0   Adam    Child       Teenager
1   Barry   Child       Adult
2   Ben     Adult       Adult
3   Adam    Teenager    Teenager
4   Barry   Adult       Adult
5   Ben     Baby        Adult
网页内容由stack overflow 提供, 点击上面的
可以查看英文原文，
原文链接