* Updating test matrix for 7.6 + removing oss for now.
* Resolving 7.6.0 docs issues
* Updating ML docs
* Bumping version following doc fixes
* Change ExternalMLModel to ImportedMLModel
* Refactoring of plotting + fixes for multiple charts
Updates to plotting inline with pandas 0.25.3
Enables plotting of multiple histograms on the
same figure.
* Fix to setup.py to allow submodules
+ reformat of code and better Series.hist docs
* Improved read_csv docs + made 'to_eland' params consistent
Note, will change API.
* Removing additional args from pytest.
doctests + nbval tests in the CI are not addressed by
this PR.
* Adding python 3.5 compatibility.
Main issue is ordering of dictionaries.
* Updating notebooks with 3.7 results.
* Removing tempoorary code.
* Defaulting to OrderedDict for python 3.5 + lint all code
All code reformated by PyCharm and inspection results analysed.
* Adding support for multiple arithmetic operations.
Added new 'arithmetics' file to manage this process.
More tests to be added + cleanup.
* Signficant refactor to arithmetics and mappings.
Work in progress. Tests don't pass.
* Major refactor to Mappings.
Field name mappings were stored in different places
(Mappings, QueryCompiler, Operations) and needed to
be keep in sync.
With the addition of complex arithmetic operations
this became complex and difficult to maintain. Therefore,
all field naming is now in 'FieldMappings' which
replaces 'Mappings'.
Note this commit removes the cache for some of the
mapped values and so the code is SIGNIFICANTLY
slower on large indices.
In addition, the addition of date_format to
Mappings has been removed. This again added more
unncessary complexity.
* Adding OrderedDict for 3.5 compatibility
* Fixes to ordering issues with 3.5
* Adding simple cache for mappings in flatten
Improves performance significantly on large
datasets (>10000 rows).
* Adding updated notebooks (new info_es).
All tests (doc + nbval + pytest) pass.
* Fixing issue with non-zero offset histograms.
* Adding python 3.5 compatibility.
Main issue is ordering of dictionaries.
* Updating notebooks with 3.7 results.
* Removing tempoorary code.
* Defaulting to OrderedDict for python 3.5 + lint all code
All code reformated by PyCharm and inspection results analysed.
* Adding support for multiple arithmetic operations.
Added new 'arithmetics' file to manage this process.
More tests to be added + cleanup.
* Signficant refactor to arithmetics and mappings.
Work in progress. Tests don't pass.
* Major refactor to Mappings.
Field name mappings were stored in different places
(Mappings, QueryCompiler, Operations) and needed to
be keep in sync.
With the addition of complex arithmetic operations
this became complex and difficult to maintain. Therefore,
all field naming is now in 'FieldMappings' which
replaces 'Mappings'.
Note this commit removes the cache for some of the
mapped values and so the code is SIGNIFICANTLY
slower on large indices.
In addition, the addition of date_format to
Mappings has been removed. This again added more
unncessary complexity.
* Adding OrderedDict for 3.5 compatibility
* Fixes to ordering issues with 3.5
* Adding simple cache for mappings in flatten
Improves performance significantly on large
datasets (>10000 rows).
* Adding updated notebooks (new info_es).
All tests (doc + nbval + pytest) pass.
* Adding python 3.5 compatibility.
Main issue is ordering of dictionaries.
* Updating notebooks with 3.7 results.
* Removing tempoorary code.
* Defaulting to OrderedDict for python 3.5 + lint all code
All code reformated by PyCharm and inspection results analysed.
* Adding support for multiple arithmetic operations.
Added new 'arithmetics' file to manage this process.
More tests to be added + cleanup.
* Signficant refactor to arithmetics and mappings.
Work in progress. Tests don't pass.
* Major refactor to Mappings.
Field name mappings were stored in different places
(Mappings, QueryCompiler, Operations) and needed to
be keep in sync.
With the addition of complex arithmetic operations
this became complex and difficult to maintain. Therefore,
all field naming is now in 'FieldMappings' which
replaces 'Mappings'.
Note this commit removes the cache for some of the
mapped values and so the code is SIGNIFICANTLY
slower on large indices.
In addition, the addition of date_format to
Mappings has been removed. This again added more
unncessary complexity.
* Adding OrderedDict for 3.5 compatibility
* Fixes to ordering issues with 3.5
