如何组合不同大小的数据框......？

Question

我正在尝试将项目列表合并到一个主数据框中，但我似乎无法弄清楚如何将它们合并在一起？我生成的框架大小不同，但大多数列名都是相同的，除了一两个....

所以基本上，我正在列出这样的项目阶段...（有些项目只有 2 或 3 个阶段，而其他项目有 8 或 9 个阶段..）示例：

Stage 1 SUCCESS
stage 2 SUCCESS
stage 3 SUCCESS
stage 4 DELAYED
stage 5 PENDING

并且，我在 python 循环中生成了如下所示的数据帧...

df

       project_name    Stage 1    Stage 2     
0      project 1       SUCCESS    DELAYED

df

       project_name    Stage 1    Stage 2    Stage 3    Stage 4   Stage 5 
0      project-2       NaN        NaN        NaN        NaN       NaN

df

       project_name    Stage 1    Stage 2    Stage 3    Stage 4   Stage 5   Stage 6    Stage 7   Stage 8
0      project-3       NaN        NaN        STARTED    ABANDONED NaN       NaN        NaN       
    NaN

但是，我似乎无法弄清楚如何生成包含所有其他帧的主数据帧...

# items passed in from other function...
project_data = [('Stage 1','SUCCESS'),('Stage 2','DELAYED')]
project_name = 'project-x' 
project_headers = ['Stage 1','Stage 2','Stage 3','Stage 4','Stage 5','Stage 6']
project_displayname = ''

# Create the pandas DataFrame
try:
    df
except NameError:
    print("Well, 'df' WASN'T defined after all!")
    df = pd.DataFrame( columns = project_headers, index=['0'])
else:
    df = df.reindex(list(range(0, 1))).reset_index(drop=True)
    df['project_name'] = project_name
    df.loc[df.project_name == project_name, "project"] = project_displayname


combined_frame = pd.DataFrame(columns = ['project_name']) # empty frame with one colum for merge
for details in project_data:
    (item, item_status) = details
    if item not in df:
        df[item] = np.nan
    df.loc[df.project_name == project_name, item] = item_status
    print('')
    print('')
    print(df)  
    print('')

# Which gives us a generated dataframe.... like so... 
#project_name    Stage 1    Stage 2    Stage 3    Stage 4   Stage 5   Stage 6    Stage 7   Stage 8
#project-3       NaN        NaN        STARTED    ABANDONED NaN       NaN        NaN       NaN


    #final_frame = combined_frame.merge(df, how='left')
    try:
        final_frame = pd.merge(df, combined_frame, how='outer', left_index=True, right_on=combined_frame.iloc[: , -1])
    except IndexError:
        final_frame = df.reindex_axis(df.columns.union(combined_frame.columns), axis=1)

print(final_frame)

当我运行代码时出现错误：Empty DataFrame

或者，我得到...

Columns: [project, project_name, Stage 1, Stage 2, Stage 3, Stage 4, Stage 5, Stage 6, Stage 7, Stage 8, Stage 9]
Index: []

或者我得到...

Columns: [project, project_name, Stage 1, Stage 2, Stage 3, Stage 4, Stage 5, Stage 6, Stage 7, Stage 8, Stage 9, project_x, project_name_x, Stage 1_x, Stage 2_x, Stage 3_x, Stage 4_x]
Index: []

有人能指出我方法中的错误吗？显然我错过了什么？

我想尝试获得这样的输出：

   project_name    Stage 1    Stage 2    Stage 3    Stage 4   Stage 5   Stage 6    Stage 7   Stage 8
0  project-1       STARTED    NaN        NaN        NaN       NaN       NaN        NaN       NaN
1  project-2       STARTED    STARTED    STARTED    DELAYED   NaN       NaN        NaN       NaN
2  project-3       NaN        NaN        STARTED    ABANDONED NaN       NaN        NaN       NaN
3  project-4       NaN        NaN        STARTED    ABANDONED NaN       STARTED    NaN       NaN
4  project-5       CANCELED   NaN        NaN        NaN       NaN       NaN        NaN       NaN
5  project-6       DELAYED    DELAYED    STARTED    ABANDONED NaN       NaN        STARTED    NaN

提前致谢，

E

Answer 1

您可以根据输入数据轻松构建单独的框架：

# items passed in from other function...
project_data = [('Stage 1','SUCCESS'),('Stage 2','DELAYED')]
project_name = 'project-x' 
project_headers = ['Stage 1','Stage 2','Stage 3','Stage 4','Stage 5','Stage 6']
project_displayname = ''

df = pd.DataFrame([dict(project_data)], columns = ['project','project_name']
                  + project_headers)
df.loc[:, ['project', 'project_name']] = [[project_name, project_displayname]]

它将给 df:

     project project_name  Stage 1  Stage 2  Stage 3  Stage 4  Stage 5  Stage 6
0  project-x               SUCCESS  DELAYED      NaN      NaN      NaN      NaN

然后您可以使用 pd.concat 连接所有单独的数据帧。唯一的限制是您必须事先知道所有列的名称（或此处的最大阶段数...）

如何组合不同大小的数据框......？

How does one combine a dataframe of different sizes....?

python

dataframe

merge