熊猫-基于值复制行

您可以使用:

• 第一次重塑 set_index 和 unstack 对于 Series Multiindex
• repeat 按重复所有行的索引 loc
• reset_index 用于转换 MultiIndex 到列
• 删除列 0 通过 drop
• 重命名列并更改顺序 reindex

s = df.set_index(['Company','WEEK_DAYS','value']).stack()
df = (s.loc[s.index.repeat(s)]
       .reset_index()
       .drop(0, axis=1)
       .rename(columns={'level_3':'WEEK'})
       .reindex(columns=['Company','WEEK_DAYS','WEEK','value'])
      )
print (df)
  Company WEEK_DAYS          WEEK  value
0  google   MON-FRI  SEPTEMBER 29    0.5
1  google   MON-FRI  SEPTEMBER 29    0.5
2  google   MON-FRI  SEPTEMBER 29    0.5
3  google   MON-FRI  SEPTEMBER 29    0.5
4  google   MON-FRI  SEPTEMBER 29    0.5
5  google       TUE  SEPTEMBER 15    0.7
6  google       TUE  SEPTEMBER 15    0.7
7  google       TUE  SEPTEMBER 15    0.7
8  google       TUE  SEPTEMBER 22    0.7
9  google       TUE  SEPTEMBER 22    0.7

你可以利用 pandas.DataFrame.from_records 作用

我建议您在新的数据框架中创建一个值列表,即

records = [('google', 'MON-FRI', 'SEPTEMBER 29', '0.5')]
# assumes only one record for MON-FRI .. you might need to do some handling here otherwise
records *= int(df['SEPTEMBER 29'][df[WEEK_DAYS] == 'MON-FRI'])
labels = ['Company', 'WEEK_DAYS', 'WEEK', 'value']

new_df = pd.DataFrame.from_records(records, columns=labels)

我想这不是有史以来最优雅的解决方案,但应该会奏效。