Question

我正在使用pandasdmx版本0.7.0，虽然我已经成功使用频率维度的OECD数据集，但还有其他数据集，例如Fossil Fuel支持数据集，其中没有频率维度我无法创建一个pandas DataFrame，如下所示：

from pandasdmx import Request

# http://stats.oecd.org/sdmx-json/data/FFS_BRA/all/all

oecd = Request('OECD')

try:
    data_response = oecd.data(resource_id='FFS_BRA', key='all/all')
except UnicodeDecodeError:
    pass
except KeyError:
    pass
else:
    oecd_data = data_response.data

    print(oecd_data.dim_at_obs)
    series_list = list(oecd_data.series)
    print(len(series_list))
    print(series_list[0].key)
    print(set(s.key.FUEL for s in oecd_data.series))

    fuel = (s for s in oecd_data.series if s.key.FUEL == 'HARDCOAL')
    df = data_response.write(fuel)

print("completed ...")

Output:

TIME_PERIOD
25
SeriesKeyTuple(MEA='BRA_DT_03', FUEL='HARDCOAL',  IND='CSE',INC='CONSUMPTION', STG='GENER', MEC='DT', LVL='FED')
{'NONBIOJETK', 'LIGNITE', 'HARDCOAL', 'NAPHTHA', 'NONBIODIES', 'CRUDEOIL', 'RESFUEL', 'NATGAS', 'NGL', 'LPG', 'NONBIOGASO'}

File "stackoverflow_dataframe.py", line 23, in <module>
df = data_response.write(fuel)
File "pandasdmx\api.py", line 635, in write
return self._writer.write(source=source, **kwargs)
File "pandasdmx\writer\data2pandas.py", line 109, in write
reverse_obs, fromfreq, parse_time))
File "pandasdmx\writer\data2pandas.py", line 107, in <genexpr>
series_list = list(s for s in self.iter_pd_series(
File "pandasdmx\writer\data2pandas.py", line 228, in iter_pd_series
obs_values, index=series_index, name=series.key)
UnboundLocalError: local variable 'series_index' referenced before assignment

这是非频率维度数据集的正确方法，还是我刚编错了？

Answer 1

pandasdmx（0.7.0）作者的正式回答是使用写入函数的parse_time参数设置为False。然后，write函数不会尝试生成DateTime索引，而是获取字符串索引。

CREATE TABLE `tbl_usr_login` (
      `LoginID` int(11) NOT NULL AUTO_INCREMENT,
      `UID` int(50) NOT NULL,
      `ip_address` varchar(55) DEFAULT NULL,
      `device` varchar(100) DEFAULT NULL,
      `time_stamp` datetime DEFAULT NULL,
      PRIMARY KEY (`LoginID`)
    )ENGINE=InnoDB DEFAULT CHARSET=latin1;

由于pandasdmx使用SDMX ID为列名和数据返回一个宽格式表，并且由于时间压力，我编写了自己的表来返回一个长格式表（我可以根据需要进行调整），并可以选择使用SDMX ID或完整描述。该解决方案已在835个经合组织数据集上进行了测试！

from pandasdmx import Request

# http://stats.oecd.org/sdmx-json/data/FFS_BRA/all/all

oecd = Request('OECD')

try:
    data_response = oecd.data(resource_id='FFS_BRA', key='all/all')
except UnicodeDecodeError:
     pass
except KeyError:
    pass
else:
    oecd_data = data_response.data

    df = data_response.write(oecd_data.series, parse_time=False)
    df.to_csv('FFS_BRA_solution1.csv')

pandasdmx无法为某些OECD数据创建DataFrame

1 个答案: