[release] 0.7.1 🎉 (#94)

* [patch:lib] Rework preprocessing module (#75)

* Rework sequentia.preprocessing

* Remove Test.ipynb

* [patch:docs] Update README.md (#76)

* [patch:pkg] Clean up package imports (#77)

* [release] 0.7.0a1 🎉 (#78)

* [patch:ci] Fix pomegranate to v0.12.0 (#79)

* [add:lib] Add (de)serialization support for all classifiers (#80)

* Finish tests for HMM and HMMClassifier

* Finish serialization documentation

* Ensure no NaNs in tests

* Ensure Nans in HMM.test_as_dict_with_nan

* Ensure NaNs in HMM.test_as_dict_with_nan

* Ensure NaNs in HMM.test_as_dict_with_nan

* Remove HMM NaN serialization test

* [patch:docs] Finish preprocessing documentation and tests (#81)

* [patch:lib] Simplify package imports (#82)

* [patch:docs] Fix HMM topology documentation typo (#83)

* [patch:lib] Remove nested helper functions in DTWKNN.predict() (#84)

* [add:lib] Add strict left-right HMM topology (#85)

* [patch:lib] Fix strict left-right topology serialization (#86)

* [add:lib] Implement GMM-HMMs (#87)

* Implement GMMHMMs

* Fix strict left-right topology serialization for GMMHMM

* Finish GMMHMM documentation

* Finish GMMHMM documentation

* Fix GMMHMM.as_dict method reference typo

* Validate stored models before deserialization

* Finish GMMHMM tests

* Update README

* Attempt to fix randomness in HMM and GMMHMM tests

* [add:lib] Implement custom, uniform and frequency-based HMM priors (#88)

* Implement custom HMM classification priors

* Fix custom priors for MAP classification

* Fix HMMClassifier specs

* Move README image

* Fix randomness in HMMClassifier tests

* [patch:tests] Fix GMMHMM NaN errors in tests (#89)

* [add:lib] Implement distance-weighted DTW-kNN predictions (#90)

* Implement distance-weighted DTWKNN

* Update README.md

* [patch:lib] Rename DTWKNN to KNNClassifier (#91)

* Rename DTWKNN to KNNClassifier

* Fix HMM test_load_dict

* [patch:lib] Fix issues with KNNClassifier deserialization (#93)

* Release v0.7.1

CHANGELOG.md

@@ -1,3 +1,11 @@
+## [0.7.1](https://github.com/eonu/sequentia/releases/tag/v0.7.1)
+
+#### Major changes
+
+- Fix deserialization for `KNNClassifier`. ([#93](https://github.com/eonu/sequentia/pull/93))
+  - Sort HDF5 keys before loading as `numpy.ndarray`s.
+  - Pass `weighting` function into deserialization constructor.
+
 ## [0.7.0](https://github.com/eonu/sequentia/releases/tag/v0.7.0)
 
 #### Major changes

docs/conf.py

@@ -24,7 +24,7 @@
 author = 'Edwin Onuonga'
 
 # The full version, including alpha/beta/rc tags
-release = '0.7.0'
+release = '0.7.1'
 
 # -- General configuration ---------------------------------------------------
 

lib/sequentia/classifiers/knn/knn_classifier.py

@@ -241,11 +241,11 @@ def load(cls, path, encoding='utf-8', metric=euclidean, weighting=(lambda x: 1))
         with h5py.File(path, 'r') as f:
             # Deserialize the model hyper-parameters
             params = f['params']
-            clf = cls(k=int(params['k'][()]), radius=int(params['radius'][()]), metric=metric)
+            clf = cls(k=int(params['k'][()]), radius=int(params['radius'][()]), metric=metric, weighting=weighting)
 
             # Deserialize the training data and labels
             X, y = f['data']['X'], f['data']['y']
-            clf._X = [np.array(X[k]) for k in X.keys()]
+            clf._X = [np.array(X[k]) for k in sorted(X.keys(), key=lambda k: int(k))]
             clf._y = [label.decode(encoding) for label in y]
 
         return clf

setup.py

@@ -3,7 +3,7 @@
 from __future__ import print_function
 from setuptools import setup, find_packages
 
-VERSION = '0.7.0'
+VERSION = '0.7.1'
 
 with open('README.md', 'r') as fh:
     long_description = fh.read()