用dask合并数据框并将其转换为熊猫

Question

我有两个数据框

dataframe1：

    >df_case = dd.read_csv('s3://../.../df_case.csv')
    >df_case.head(1)
sacc_id$                   id$             creation_date
 0  001A000000hwvV0IAI  5001200000ZnfUgAAJ  2016-06-07 14:38:02

dataframe2：

>df_limdata = dd.read_csv('s3://../.../df_limdata.csv')
>df_limdata.head(1)
     sacc_id$            opp_line_id$           oppline_creation_date
0   001A000000hAUn8IAG  a0W1200000G0i3UEAR  2015-06-10

首先，我合并了两个数据框：

> case = dd.merge(df_limdata, df_case, left_on='sacc_id$',right_on='sacc_id$')

>case

Dask DataFrame Structure:
    Unnamed: 0_x    sacc_id$    opp_line_id$_x  oppline_creation_date_x     Unnamed: 0_y    opp_line_id$_y  oppline_creation_date_y
npartitions=5                           
    int64   object  object  object  int64   object  object
    ...     ...     ...     ...     ...     ...     ...
...     ...     ...     ...     ...     ...     ...     ...
    ...     ...     ...     ...     ...     ...     ...
    ...     ...     ...     ...     ...     ...     ...
Dask Name: hash-join, 78 tasks

然后我尝试将这个普通案例数据框转换为pandas数据框：

> # conversion to pandas
df = case.compute()

我收到此错误：

ValueError: Mismatched dtypes found in `pd.read_csv`/`pd.read_table`.

+------------+---------+----------+
| Column     | Found   | Expected |
+------------+---------+----------+
| Unnamed: 0 | float64 | int64    |
+------------+---------+----------+

Usually this is due to dask's dtype inference failing, and
*may* be fixed by specifying dtypes manually by adding:

dtype={'Unnamed: 0': 'float64'}

to the call to `read_csv`/`read_table`.

Alternatively, provide `assume_missing=True` to interpret
all unspecified integer columns as floats.

您能帮我解决这个问题吗？

谢谢

Answer 1

在读取文件时，dask假设列“ Unnamed：0”具有int64作为dtype，但后来在计算时发现它为float64。

因此，您在读取文件时需要提及dtype：

df_case = dd.read_csv('s3://../.../df_case.csv',dtpye={'Unnamed: 0': 'float64'})

df_limdata = dd.read_csv('s3://../.../df_limdata.csv',dtpye={'Unnamed: 0': 'float64'})


case = dd.merge(df_limdata, df_case, left_on='sacc_id$',right_on='sacc_id$')
# conversion to pandas
df = case.compute()