解决有关 dtype 是对象的 MultiIndex 错误

Question

当尝试使用 googledrive api 将旋转数据框上传到 googlesheet 时，系统提示我出现以下错误：

TypeError：不支持将类 'pandas.core.indexes.multi.MultiIndex' dtype 设置为除 object 之外的任何内容

在检查 dtypes 后，我发现我的索引值是对象，并且我也将列值转换为对象，但我仍然收到错误。

import gameData_connection
import pandas as pd
import gspread
from oauth2client.service_account import ServiceAccountCredentials
from df2gspread import df2gspread as d2g

scope = ['https://spreadsheets.google.com/feeds']
credentials = ServiceAccountCredentials.from_json_keyfile_name('credentials_files.json', scope)
gc = gspread.authorize(credentials)
spreadsheet_key = 'my_spreadsheet_key'
wks_name = 'My File Name'

connection = gameData_connection.getConnection()

print('Connect Succesful')

try:
    with connection.cursor() as cursor:
        sql1 = This is my whole sql query

cursor.execute(sql1)
        df1 = pd.read_sql_query(sql1, connection)
        print('cursor.description:', cursor.description)
        print()

        for row in cursor:
            print(row)

finally:
    connection.close()



table1 = pd.pivot_table(df1, index = ["game", "platform"], 
                        values = ["new_users", "dau", "day1_retention", "day6_retention", "revenue", "arpdau", "arppu", "pau"],
                        columns = ["year_time", "month_time"])


d2g.upload(table1, spreadsheet_key, wks_name, credentials=credentials, row_names=True) 
print(table1)

Out[21]: 
time_join           int64
year_time           int64
month_time          int64
game               object
platform           object
new_users         float64
dau               float64
day1_retention    float64
day6_retention    float64
revenue           float64
arpdau            float64
arppu             float64
pau               float64
dtype: object

我不明白我在这里做错了什么。如果我运行

d2g.upload(df1, spreadsheet_key, wks_name, credentials=credentials, row_names=True)

，它就会起作用，所以表 1 内一定发生了问题。

现在请注意，如果编写相同的 table1 来执行 excel，它就可以工作。

预先感谢您的帮助

Answer 1

我在使用 df2gspread 库实现时遇到了类似的问题。在这里提供我的解决方案，因为这个问题没有得到解答（而且很老了！）

问题

df2gspread 库的编写方式是尝试将数据帧的列和索引转换为字符串。 Pandas 不支持 MultiIndex 类型的转换，并会生成您看到的 TypeError。 问题已发布在此处的存储库中，但尚未解决。

解决方案

解决方案是使用 gspread_pandas 库，该库具有支持直接包括 MultiIndex 行和列的数据帧的实现，如示例中的数据透视表。

应用于所提供示例的解决方案

import gameData_connection
import pandas as pd
import gspread
from oauth2client.service_account import ServiceAccountCredentials
from gspread_pandas import Spread # NEW LINE

scope = ['https://spreadsheets.google.com/feeds']
credentials = ServiceAccountCredentials.from_json_keyfile_name('credentials_files.json', scope)
gc = gspread.authorize(credentials)
spreadsheet_key = 'my_spreadsheet_key'
wks_name = 'My File Name'

# NEW LINE
sh = Spread(spred=spreadsheet_key, creds=credentials)

connection = gameData_connection.getConnection()

print('Connect Successful')

try:
    with connection.cursor() as cursor:
        sql1 = This is my whole sql query

cursor.execute(sql1)
        df1 = pd.read_sql_query(sql1, connection)
        print('cursor.description:', cursor.description)
        print()

        for row in cursor:
            print(row)

finally:
    connection.close()

table1 = pd.pivot_table(df1, index = ["game", "platform"], 
                        values = ["new_users", "dau", "day1_retention", "day6_retention", "revenue", "arpdau", "arppu", "pau"],
                        columns = ["year_time", "month_time"])

# NEW LINE
sh.df_to_sheets(df=table1, # Your dataframe
                sheet=wks_name,  # Target sheet name
                start='A1', # First cell to place the data in
                index=True, # Show the index
                headers=True, # Show the columns
                merge_headers=True, # Merge the cells of multiindex headers
                merge_index=True, # Merge the rows of the muliindex index
                freeze_headers=True, # Freeze the header rows
               )