Re-write to fix references to old name, "SMITE"

.coveragerc

@@ -1,5 +1,5 @@
 [run]
-source = smite
+source = smrt
 omit =
     */python?.?/*
     */lib-python/?.?/*.py

.travis.yml

@@ -13,7 +13,7 @@ addons:
 env:
   global:
     # Directory where tests are run from
-    - TEST_DIR=/tmp/smite
+    - TEST_DIR=/tmp/smrt
     - OMP_NUM_THREADS=4
     - OPENBLAS_NUM_THREADS=4
   matrix:

README.md

@@ -4,16 +4,19 @@
 ![Supported versions](https://img.shields.io/badge/python-3.5-blue.svg) 
 
 
-# Synthetic Minority Interpolation TEchnique
-*SMOTE's younger, more sophisticated cousin*
+# SMRT
+([Original reference (enable audio)](https://www.youtube.com/watch?v=tcGQpjCztgA))
+
+##Synthetic Minority Reconstruction Technique
+*Handle your class imbalance more intelligently by using SMOTE's younger, more sophisticated cousin*
 
 
 ### Installation
 
 Installation is easy. After cloning the project onto your machine and installing the required dependencies, simply use the `setup.py` file:
 
 ```bash
-$ git clone https://github.com/tgsmith61591/smite.git
-$ cd smite
+$ git clone https://github.com/tgsmith61591/smrt.git
+$ cd smrt
 $ python setup.py install
 ```

build_tools/travis/test_script.sh

@@ -22,9 +22,9 @@ run_tests() {
     cd $TEST_DIR
 
     if [[ "$COVERAGE" == "true" ]]; then
-        nosetests -s --with-coverage --with-timer --timer-top-n 20 smite
+        nosetests -s --with-coverage --with-timer --timer-top-n 20 smrt
     else
-        nosetests -s --with-timer --timer-top-n 20 smite
+        nosetests -s --with-timer --timer-top-n 20 smrt
     fi
 
 }

setup.cfg

@@ -14,7 +14,7 @@ test = nosetests
 exe = 1
 cover-html = 1
 cover-html-dir = coverage
-cover-package = smite
+cover-package = smrt
 
 detailed-errors = 1
 with-doctest = 1

setup.py

@@ -2,12 +2,11 @@
 #
 # Author: Taylor Smith <[email protected]>
 #
-# Setup the SMITE module
+# Setup the SMRT module
 
 from __future__ import print_function, absolute_import, division
 from distutils.command.clean import clean
 from setuptools import setup
-import numpy as np
 import os
 import sys
 
@@ -17,12 +16,12 @@
     import builtins
 
 # Hacky, adopted from sklearn. This sets a global variable
-# so smite __init__ can detect if it's being loaded in the setup
+# so smrt __init__ can detect if it's being loaded in the setup
 # routine, so it won't load submodules that haven't yet been built.
-builtins.__SMITE_SETUP__ = True
+builtins.__SMRT_SETUP__ = True
 
 # metadata
-DISTNAME = 'smite'
+DISTNAME = 'smrt'
 DESCRIPTION = 'Handle class imbalance intelligently by using autoencoders ' \
               'to generate synthetic observations of your minority class.'
 
@@ -31,8 +30,8 @@
 LICENSE = 'new BSD'
 
 # import restricted version
-import smite
-VERSION = smite.__version__
+import smrt
+VERSION = smrt.__version__
 
 # get the installation requirements:
 with open('requirements.txt') as req:
@@ -43,7 +42,7 @@
 class CleanCommand(clean):
     description = "Remove build artifacts from the source tree"
 
-    # this is mostly in case we ever add a Cython module to SMITE
+    # this is mostly in case we ever add a Cython module to SMRT
     def run(self):
         clean.run(self)
         # Remove c files if we are not within a sdist package
@@ -97,7 +96,7 @@ def run(self):
                                   'Operating System :: MacOS',
                                   'Programming Language :: Python :: 2.7'
                                  ],
-                     keywords='sklearn scikit-learn sknn scikit-neuralnetwork auto-encoders class-imbalance',
+                     keywords='sklearn scikit-learn tensorflow scikit-neuralnetwork auto-encoders class-imbalance',
                      packages=[DISTNAME],
                      install_requires=REQUIREMENTS,
                      cmdclass=cmdclass)

smite/__init__.py → smrt/__init__.py

@@ -2,7 +2,7 @@
 #
 # Author: Taylor Smith <[email protected]>
 #
-# The SMITE module
+# The SMRT module
 
 import sys
 import os
@@ -13,18 +13,18 @@
     # this var is injected in the setup build to enable
     # the retrieval of the version number without actually
     # importing the un-built submodules.
-    __SMITE_SETUP__
+    __SMRT_SETUP__
 except NameError:
-    __SMITE_SETUP__ = False
+    __SMRT_SETUP__ = False
 
-if __SMITE_SETUP__:
-    sys.stderr.write('Partial import of SMITE during the build process.' + os.linesep)
+if __SMRT_SETUP__:
+    sys.stderr.write('Partial import of SMRT during the build process.' + os.linesep)
 else:
     __all__ = [
         'autoencode.py',
         'balance'
     ]
 
     # top-level imports
-    from .balance import smite_balance
+    from .balance import smrt_balance
     from .autoencode import AutoEncoder

smite/balance.py → smrt/balance.py

@@ -2,7 +2,7 @@
 #
 # Author: Taylor Smith <[email protected]>
 #
-# The SMITE balancer
+# The SMRT balancer
 
 from __future__ import division, absolute_import, division
 from numpy.random import RandomState
@@ -17,39 +17,34 @@
 import numpy as np
 
 __all__ = [
-    'smite_balance'
+    'smrt_balance'
 ]
 
 MAX_N_CLASSES = 100  # max unique classes in y
 MIN_N_SAMPLES = 2  # min n_samples per class in y
 
 
-def _validate_layers(layers):
-    if not all(isinstance(x, LayerParameters) for x in layers):
-        raise ValueError('layers should be a list, tuple or dict of smite.balance.LayerParameters')
-
-
 def _validate_ratios(ratio, name):
     if not 0. < ratio <= 1.:
         raise ValueError('Expected 0 < %s <= 1, but got %r' 
                          % (name, ratio))
 
 
-def smite_balance(X, y, return_encoders=False, balance_ratio=0.2, eps=1.0, random_state=None,
-                  parameters=None, learning_rule='sgd', learning_rate=0.01, learning_momentum=0.9, batch_size=1, 
-                  n_iter=None, n_stable=10, f_stable=0.001, valid_set=None, valid_size=0.0, normalize=None, regularize=None, 
-                  weight_decay=None, dropout_rate=None, loss_type=None, callback=None, debug=False, verbose=None, 
-                  **auto_encoder_params):
-    """SMITE (Sythetic Minority Interpolation TEchnique) is the younger, more sophisticated cousin to
-    SMOTE (Synthetic Minority Oversampling TEchnique). Using auto-encoders, SMITE learns the parameters 
+def smrt_balance(X, y, return_encoders=False, balance_ratio=0.2, eps=1.0, random_state=None,
+                 parameters=None, learning_rule='sgd', learning_rate=0.01, learning_momentum=0.9, batch_size=1,
+                 n_iter=None, n_stable=10, f_stable=0.001, valid_set=None, valid_size=0.0, normalize=None, regularize=None,
+                 weight_decay=None, dropout_rate=None, loss_type=None, callback=None, debug=False, verbose=None,
+                 **auto_encoder_params):
+    """SMRT (Sythetic Minority Reconstruction Technique) is the younger, more sophisticated cousin to
+    SMOTE (Synthetic Minority Oversampling TEchnique). Using auto-encoders, SMRT learns the parameters
     that best reconstruct the observations in each minority class, and then generates synthetic observations
     until the minority class is represented at a minimum of ``balance_ratio`` * majority_class_size. 
 
-    SMITE avoids one of SMOTE's greatest risks: In SMOTE, when drawing random observations from whose k-nearest 
+    SMRT avoids one of SMOTE's greatest risks: In SMOTE, when drawing random observations from whose k-nearest
     neighbors to reconstruct, the possibility exists that a "border point," or an observation very close to 
     the decision boundary may be selected. This could result in the synthetically-generated observations lying 
     too close to the decision boundary for reliable classification, and could lead to the degraded performance
-    of an estimator. SMITE avoids this risk, by ranking observations according to their reconstruction MSE, and
+    of an estimator. SMRT avoids this risk, by ranking observations according to their reconstruction MSE, and
     drawing samples to reconstruct from the lowest-MSE observations (i.e., the most "phenotypical" of a class).
     
     Parameters
@@ -233,14 +228,14 @@ def smite_balance(X, y, return_encoders=False, balance_ratio=0.2, eps=1.0, rando
     y_type = type_of_target(y)
     supported_types = {'multiclass', 'binary'}
     if y_type not in supported_types:
-        raise ValueError('SMITE balancer only supports %r, but got %r' % (supported_types, y_type))
+        raise ValueError('SMRT balancer only supports %r, but got %r' % (supported_types, y_type))
 
     present_classes, counts = np.unique(y, return_counts=True)
     n_classes = len(present_classes)
 
     # ensure <= MAX_N_CLASSES
     if n_classes > MAX_N_CLASSES:
-        raise ValueError('SMITE balancer currently only supports a maximum of %i '
+        raise ValueError('SMRT balancer currently only supports a maximum of %i '
                          'unique class labels, but %i were identified.' % (MAX_N_CLASSES, n_classes))
 
     # check layers:

smite/tests/test_smite.py → smrt/tests/test_smrt.py

@@ -1,6 +1,6 @@
 
 from __future__ import division, absolute_import, division
-from smite.balance import smite_balance
+from smrt.balance import smrt_balance
 from sklearn.datasets import load_iris
 
 
@@ -11,5 +11,5 @@ def test_iris():
     # undersample 1, 2
     indices = [i for i in range(50)] + [i for i in range(50,60)] + [i for i in range(100, 120)]
     X, y = X[indices, :], y[indices]
-    X, y, encoders = smite_balance(X, y, return_encoders=True, balance_ratio=1.0)
+    X, y, encoders = smrt_balance(X, y, return_encoders=True, balance_ratio=1.0)
     assert X.shape[0] == 150, 'Expected 150, got %i' % X.shape[0]