CHANGES
=======

0.7.2
-----

* Update requirements for python 2

0.7.1
-----

* fix print issue when categories are not strings
* fix typo
* give the ability to multi\_model to predict when trained to one categorical feature (constant) that do not exist in future data test that when we dont have the constant categorical feature in the future the model still uses the one existing model and predicts
* # convert notebook example

0.6.9
-----

* change dtype of prediction to float64

0.6.8
-----

* add different weights in the objectives function for "under\_predict\_weight"

0.6.7
-----

* add package version retrieval
* python2 compatible
* add requirements
* pep changes
* pep changes
* pep changes
* pep changes
* Able to have different categorical encoders for, nominal and for ordinal categorical features
* Able to have different categorical encoders for, nominal and for ordinal categorical features
* add tests
* add comments of dipanjan in the code
* add comments of dipanjan in the code
* complete experiment with 2 transformers and saving the model
* add a new transformatiom
* make changes of how to import modules
* make changes of how to import modules
* add grouped models, transformers and a jupyter example

0.6.6
-----

* fix predict for unseend data that do not belong to a category
* change tests to be compatible with the new modules
* exchanging the previous feature\_subset classes with the ones made in the branch
* add minor changes to make the notebook problem specific
* add minor changes to make the notebook problem specific
* add minor changes to make the notebook problem specific
* complete feature\_subset\_example2
* complete feature\_subset\_example2
* change subset model and create a new example
* add tests for prediction model preprocessing and for feature\_subset

0.6.5
-----

* Fix twine command

0.6.4
-----

* Fix tag build

0.6.3
-----

* Add PBR supported setup scheme and circleci based release
* Add tests for prediction model
* Add pytest dependency specific to python version
* Add test deps
* Install module
* Fix typo
* Add circleci config
* path changes to make the notebook reproducible
* path changes to make the notebook reproducible
* Update notebook with references to classes and rename MyTransformer
* add the jupyter notebook
* Adds a class transformer that takes data, excludes features that are used for grouping and perform every sklearn transformer. Adds a jupyter notebook that tests the transformer by itself and with a combination of the FeatureSubsetModel using pipeline and Cross Validation
* Added example of feature subset model
* Fix formatting
* Add version and release notes
* Add custom objective function
* Pin exact versions for Python 2.7
* Version gate numpy, scipy, pandas, matplotlib, ipython, scikit-learn, xgboost.Remove unused packages: jupyterlab, lightgbm and bottleneck
* Upgrade to Python 3
* Handle low-dimensional models
* library updates
* only keep features
* resolve post processing bug
* explicit train\_mode
* explicit train\_mode
* train test sets
* Silent Mode
* Silent Mode
* Version 0.5.0 including one hot encoding
* Version 0.5.0 including one hot encoding
* Version 0.5.0 including one hot encoding
* Library versions
* Library versions
* Codacy
* GBM Requirements
* Python 3 fixes
* Fixing python 3 compatibility using 2to3 fixer unicode
* Fixing python 3 compatibility using 2to3 fixer dict
* Move to Jupyter Lab & Seaborne + viz scripts
* Move to Jupyter Lab & Seaborne + viz scripts
* Improved garbage collection avoidance
* Updated mask reversal with tilde call
* Updated requirements
* Updated requirements
* Updated requirements
* Updated requirements
* Updated requirements
* Updated requirements
* Version 0.3.0 -> Cleaned up code
* Version 0.2.0
* Extended standard procedures
* improved output
* typo
* Fixed Manifest for real (I hope)
* Fixed Manifest
* Fixed Manifest
* Typo and visualizations
* visualizations
* data checks
* setup functions
* msgpack
* basic setup
* Initial commit
