Description of problem: The current version in epel7 is 0.17.1 and there are new features that have been released in the last 15 months.
There are API and backwards incompatible changes since 0.17.1: http://pandas-docs.github.io/pandas-docs-travis/ EPEL strives for stability so I'm hesitant to update.
How about 0.18.1? That added the Select by Callable that I was hoping to leverage: http://pandas.pydata.org/pandas-docs/stable/indexing.html#selection-by-callable
These are the incompatible changes: - 0.18 - http://pandas-docs.github.io/pandas-docs-travis/whatsnew.html#whatsnew-0180-api-breaking the leading whitespaces have been removed from the output of .to_string(index=False) method (GH11833) the out parameter has been removed from the Series.round() method. (GH11763) DataFrame.round() leaves non-numeric columns unchanged in its return, rather than raises. (GH11885) DataFrame.head(0) and DataFrame.tail(0) return empty frames, rather than self. (GH11937) Series.head(0) and Series.tail(0) return empty series, rather than self. (GH11937) to_msgpack and read_msgpack encoding now defaults to 'utf-8'. (GH12170) the order of keyword arguments to text file parsing functions (.read_csv(), .read_table(), .read_fwf()) changed to group related arguments. (GH11555) NaTType.isoformat now returns the string 'NaT to allow the result to be passed to the constructor of Timestamp. (GH12300) - 0.18.1 - http://pandas-docs.github.io/pandas-docs-travis/whatsnew.html#whatsnew-0181-api .groupby(..).nth() changes numpy function compatibility Using .apply on groupby resampling Changes in read_csv exceptions to_datetime error changes Other API changes Deprecations My gut instinct is to not upgrade.