Merge 55f9cad into 4909830

.travis.yml

@@ -2,10 +2,14 @@
 # check if .travis.yml is valid: http://lint.travis-ci.org/
 # to skip build for given commit put [ci skip] or [skip ci] in commit message
 
+# required for Python >= 3.7
+dist: xenial
+
 language: python
 
-# versions supported by scikit-learn
+# versions supported by scikit-learn and some additional versions
 python:
+  - "3.7"
   - "3.6"
   - "3.5"
   - "3.4"
@@ -16,6 +20,7 @@ branches:
   only:
   - master
   - dev
+  - py2
 
 install:
   - pip install numpy

CHANGELOG.md

@@ -0,0 +1,53 @@
+# Changelog
+
+### v0.4.0 -- August 12, 2019
+
+Since v0.4.0 vecstack provides official support for Python 3.5 and higher only,  
+but still there is unofficial support for Python 2.7 and Python 3.4.  
+Please see [details](https://github.com/vecxoz/vecstack/blob/master/PY2.md).
+
+Scikit-learn API:
+* Fixed #31. `sklearn.externals.six` deprecation
+* Fixed #29. Out-of-memory in `np.random.choice` for very large ranges
+
+Functional API:
+* Feature #18. Added support for N-dimensional input. Useful for convolutional nets.
+* Added aliases for `mode` parameter values which correspond to respective `variant` parameter values of `StackingTransformer`:
+  * 'oof_pred_bag' == 'A'
+  * 'oof_pred' == 'B'
+
+### v0.3.0 -- April 6, 2018
+
+Introducing Scikit-learn API: `StackingTransformer`
+
+* Standard transformer class with `fit` and `transform` methods
+* Compatible with `Pipeline` and `FeatureUnion`
+
+### v0.2.2 -- February 23, 2018
+
+* Fixed #5. Wrong behavior during sparse matrix processing
+* Improved input data validation
+* Improved sparse matrix processing
+
+### v0.2.1 -- January 24, 2018 -- Maintenance release
+
+* Minor modifications
+
+### v0.2 -- January 23, 2018
+
+New features:
+
+* Classification with probabilities
+* Modes: compute only what you need (only OOF, only predictions, both, etc.)
+* Save resulting arrays and log with model parameters
+
+### v0.1 -- November 22, 2016 -- Initial release
+
+Features:
+
+* Functional stacking API
+* Regression
+* Classification with class labels
+* Ordinary and stratified k-fold split
+* User-defined metric
+* User-defined transformations for target and prediction

LICENSE.txt

@@ -1,7 +1,7 @@
 MIT License
 
 Vecstack. Python package for stacking (machine learning technique)
-Copyright (c) 2016-2018 Igor Ivanov
+Copyright (c) 2016-2019 Igor Ivanov
 Email: vecxoz@gmail.com
 
 Permission is hereby granted, free of charge, to any person obtaining a copy

PY2.md

@@ -0,0 +1,23 @@
+### Python 3.x
+
+Since v0.4.0 vecstack provides official support for Python 3.5 and higher only,  
+but still there is unofficial support for Python 2.7 and Python 3.4. See details below.  
+
+The reason for these changes is global movement in Python 3.x direction.  
+Vecstack depends on scikit-learn which has already stopped support for Python < 3.5.  
+Scikit-learn v0.20.x is the last version supporting Python 2.7 and Python 3.4.  
+Vecstack follows this direction as well.  
+Please see [python3statement.org](https://python3statement.org/) for more details.  
+
+### Unofficial support for Python 2.7 and Python 3.4
+
+You can still install and run latest vecstack on Python 2.7 and Python 3.4.  
+NOTE. It will require legacy versions of the following packages:   
+* numpy<1.17
+* scipy<1.3
+* scikit-learn>=0.18,<0.21
+
+There is a dedicated branch on GitHub called `py2` with appropriate requirements in `setup.py`.  
+Installation:  
+
+`pip install https://github.com/vecxoz/vecstack/archive/py2.zip`

README.md

@@ -22,10 +22,10 @@ Convenient way to automate OOF computation, prediction and bagging using any num
     * Predict [class labels or probabilities](https://github.com/vecxoz/vecstack/blob/master/vecstack/coresk.py#L119) in classification task
     * Apply any [user-defined metric](https://github.com/vecxoz/vecstack/blob/master/vecstack/coresk.py#L124)
     * Apply any [user-defined transformations](https://github.com/vecxoz/vecstack/blob/master/vecstack/coresk.py#L87) for target and prediction
-    * Python 2, Python 3
+    * Python 3.5 and higher, [unofficial support for Python 2.7 and 3.4](https://github.com/vecxoz/vecstack/blob/master/PY2.md)
     * Win, Linux, Mac
     * [MIT license](https://github.com/vecxoz/vecstack/blob/master/LICENSE.txt)
-    * Depends on **numpy**, **scipy**, **scikit-learn>=18.0**
+    * Depends on **numpy**, **scipy**, **scikit-learn>=0.18**
 
 # Get started
 * [FAQ](https://github.com/vecxoz/vecstack#stacking-faq)
@@ -292,14 +292,15 @@ Stacking API comparison:
 | Estimator implementation restrictions | Must have only `fit` and `predict` (`predict_proba`) methods | Must be fully scikit-learn compatible |
 | `NaN` and `inf` in input data | Allowed | Not allowed |
 | Can automatically save OOF and log in files | Yes | No |
+| Input dimensionality (`X_train`, `X_test`) | Arbitrary | 2-D |
 
 ### 21. How do parameters of `stacking` function and `StackingTransformer` correspond?
 
-| **stacking function**   | **StackingTransformer**           |
-|-------------------------|-----------------------------------|
-| `models=[Ridge()]`      | `estimators=[('ridge', Ridge())]` |
-| `mode='oof_pred_bag'`   | `variant='A'`                     |
-| `mode='oof_pred'`       | `variant='B'`                     |
+| **stacking function**                 | **StackingTransformer**           |
+|---------------------------------------|-----------------------------------|
+| `models=[Ridge()]`                    | `estimators=[('ridge', Ridge())]` |
+| `mode='oof_pred_bag'` (alias `'A'`)   | `variant='A'`                     |
+| `mode='oof_pred'` (alias `'B'`)       | `variant='B'`                     |
 
 ### 22. Why Scikit-learn API was implemented as transformer and not predictor?
 

examples/04_sklearn_api_regression_pipeline.ipynb

@@ -512,7 +512,7 @@
    "source": [
     "# 2. Pipeline\n",
     "\n",
-    "StackingTransformer is fully scikit-learn compatible so we can easily implement **arbitrary number of stacking layers** using Pipeline\n"
+    "StackingTransformer is fully scikit-learn compatible so we can easily implement **arbitrary number of stacking levels** using Pipeline\n"
    ]
   },
   {
@@ -535,7 +535,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# If we have several stacking layers our Pipeline steps would be:\n",
+    "# If we have several stacking levels our Pipeline steps would be:\n",
     "# steps = [('stack_L1', stack_L1),\n",
     "#          ('stack_L2', stack_L2),\n",
     "#          ('stack_L99', stack_L99), # :-)\n",

setup.py

@@ -2,23 +2,26 @@
 
 from setuptools import setup
 
+long_desc = '''
+Python package for stacking (stacked generalization) featuring lightweight functional API and fully compatible scikit-learn API.
+Convenient way to automate OOF computation, prediction and bagging using any number of models.
+'''
+
 setup(name='vecstack',
-      version='0.3.0',
+      version='0.4.0',
       description='Python package for stacking (machine learning technique)',
-      long_description='Convenient way to automate OOF computation, prediction and bagging using any number of models',
+      long_description=long_desc,
       classifiers=[
           'License :: OSI Approved :: MIT License',
           'Operating System :: MacOS',
           'Operating System :: Microsoft :: Windows',
           'Operating System :: POSIX',
           'Operating System :: Unix',
           'Programming Language :: Python',
-          'Programming Language :: Python :: 2',
-          'Programming Language :: Python :: 2.7',
           'Programming Language :: Python :: 3',
-          'Programming Language :: Python :: 3.4',
           'Programming Language :: Python :: 3.5',
           'Programming Language :: Python :: 3.6',
+          'Programming Language :: Python :: 3.7',
           'Topic :: Scientific/Engineering',
           'Topic :: Scientific/Engineering :: Artificial Intelligence',
           'Topic :: Scientific/Engineering :: Information Analysis',