[ https://issues.apache.org/jira/browse/ARROW-2037?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16858692#comment-16858692 ]
Joris Van den Bossche commented on ARROW-2037: ---------------------------------------------- You get an "empty" inferred type if you have an object dtype with no rows or only missing values: {code:python} In [43]: s = pd.Series([], dtype=object) In [44]: pd.api.types.infer_dtype(s, skipna=True) Out[44]: 'empty' In [45]: s = pd.Series([None, None], dtype=object) In [46]: pd.api.types.infer_dtype(s, skipna=True) Out[46]: 'empty' {code} Not sure if that is what was meant, but the roundtrip for that is currently working, so can add a test for that. > [Python]: Add tests for ARROW-1941 cases where pandas inferred type is 'empty' > ------------------------------------------------------------------------------ > > Key: ARROW-2037 > URL: https://issues.apache.org/jira/browse/ARROW-2037 > Project: Apache Arrow > Issue Type: Bug > Components: Python > Affects Versions: 0.8.0 > Reporter: Phillip Cloud > Assignee: Phillip Cloud > Priority: Major > Fix For: 0.14.0 > > -- This message was sent by Atlassian JIRA (v7.6.3#76005)