PEP 8 compliance and tsfresh version compatibility fix

Author: prasi372

sciope/features/feature_extraction.py

@@ -71,7 +71,8 @@ def _wrapper(data):
         return total
 
     res = np.array(_wrapper(data))
-    return res.reshape(-1, res.shape[1]*res.shape[2])
+    return res.reshape(-1, res.shape[1] * res.shape[2])
+
 
 def _get_tsfresh_features_names(features):
     """
@@ -93,11 +94,10 @@ def _get_tsfresh_features_names(features):
         shape: Nr of total features
     """
     f_names = []
-    
+
     for key in features.keys():
         assert hasattr(feature_calculators, key), "%s does not exist as a feature supported by tsfresh" % key
 
-
     for function_name, parameter_list in features.items():
         func = getattr(feature_calculators, function_name)
         if parameter_list:
@@ -108,7 +108,8 @@ def _get_tsfresh_features_names(features):
 
     return f_names
 
-def remove_nan_features(x, features):   # pragma: no cover
+
+def remove_nan_features(x, features):  # pragma: no cover
     """
     Method to remove features containing NaN values.
 

sciope/inference/smc_abc.py

@@ -30,10 +30,11 @@
 from dask.distributed import futures_of, as_completed, wait
 from dask import delayed
 
+
 class PerturbationPrior(PriorBase):
 
     def __init__(self, ref_prior, samples, normalized_weights, perturbation_kernel,
-                 use_logger = False):
+                 use_logger=False):
 
         self.name = 'Perturbation Prior'
         self.ref_prior = ref_prior
@@ -42,7 +43,7 @@ def __init__(self, ref_prior, samples, normalized_weights, perturbation_kernel,
         self.perturbation_kernel = perturbation_kernel
         super(PerturbationPrior, self).__init__(self.name, use_logger)
 
-    def draw(self, n = 1, chunk_size = 1):
+    def draw(self, n=1, chunk_size=1):
 
         assert n >= chunk_size
 
@@ -57,9 +58,9 @@ def draw(self, n = 1, chunk_size = 1):
         return generated_samples
 
     @delayed
-    def _weighted_draw_perturb(self,m):
+    def _weighted_draw_perturb(self, m):
         idxs = np.random.choice(self.samples.shape[0], m,
-                                p = self.normalized_weights)
+                                p=self.normalized_weights)
         s0 = [self.samples[idx] for idx in idxs]
         s = []
         for z in s0:
@@ -72,6 +73,7 @@ def _weighted_draw_perturb(self,m):
 
         return np.asarray(s)
 
+
 class SMCABC(InferenceBase):
     """
     SMC - Approximate Bayesian Computation
@@ -145,19 +147,20 @@ def infer(self, num_samples, batch_size, chunk_size=10, ensemble_size=1, normali
 
         # Generate an initial population from the first epsilon
         abc_instance = abc_inference.ABC(self.data, self.sim, prior_function,
-                                         epsilon = self.epsilons[0],
-                                         summaries_function = self.summaries_function,
-                                         distance_function = self.distance_function,
-                                         summaries_divisor = self.summaries_divisor,
-                                         use_logger = self.use_logger)
+                                         epsilon=self.epsilons[0],
+                                         summaries_function=self.summaries_function,
+                                         distance_function=self.distance_function,
+                                         summaries_divisor=self.summaries_divisor,
+                                         use_logger=self.use_logger)
 
         print("Starting epsilon={}".format(self.epsilons[0]))
-        abc_instance.compute_fixed_mean(chunk_size = chunk_size)
-        abc_results = abc_instance.infer(num_samples = t, batch_size = batch_size, chunk_size = chunk_size, normalize = normalize)
+        abc_instance.compute_fixed_mean(chunk_size=chunk_size)
+        abc_results = abc_instance.infer(num_samples=t, batch_size=batch_size, chunk_size=chunk_size,
+                                         normalize=normalize)
 
         final_results = abc_results
         population = np.vstack(abc_results['accepted_samples'])[:t]
-        normalized_weights = np.ones(t)/t
+        normalized_weights = np.ones(t) / t
         d = population.shape[1]
 
         # SMC iterations
@@ -175,23 +178,23 @@ def infer(self, num_samples, batch_size, chunk_size=10, ensemble_size=1, normali
             try:
                 # Run ABC on the next epsilon using the proposal prior
                 abc_instance = abc_inference.ABC(self.data, self.sim, new_prior,
-                                        epsilon = eps, summaries_function = self.summaries_function,
-                                        distance_function = self.distance_function,
-                                        summaries_divisor = self.summaries_divisor,
-                                        use_logger = self.use_logger)
-                abc_instance.compute_fixed_mean(chunk_size = chunk_size)
-                abc_results = abc_instance.infer(num_samples = t,
-                                                 batch_size = batch_size,
-                                                 chunk_size = chunk_size,
-                                                 normalize = normalize)
+                                                 epsilon=eps, summaries_function=self.summaries_function,
+                                                 distance_function=self.distance_function,
+                                                 summaries_divisor=self.summaries_divisor,
+                                                 use_logger=self.use_logger)
+                abc_instance.compute_fixed_mean(chunk_size=chunk_size)
+                abc_results = abc_instance.infer(num_samples=t,
+                                                 batch_size=batch_size,
+                                                 chunk_size=chunk_size,
+                                                 normalize=normalize)
                 new_samples = np.vstack(abc_results['accepted_samples'])[:t]
 
                 # Compute importance weights for the new samples
                 prior_weights = self.prior_function.pdf(new_samples)
                 kweights = self.perturbation_kernel.pdf(population, new_samples)
 
-                new_weights = prior_weights / np.sum(kweights * normalized_weights[:, np.newaxis], axis = 0)
-                new_weights = new_weights/sum(new_weights)
+                new_weights = prior_weights / np.sum(kweights * normalized_weights[:, np.newaxis], axis=0)
+                new_weights = new_weights / sum(new_weights)
 
                 population = new_samples
                 normalized_weights = new_weights

sciope/tests/test_features.py

@@ -5,7 +5,7 @@
 
 
 def test_generate_tsfresh_features():
-    X = np.random.randn(2, 2, 100)
+    x = np.random.randn(2, 2, 100)
     features = EfficientFCParameters()
-    test = generate_tsfresh_features(X, features)
+    test = generate_tsfresh_features(x, features)
     assert test.shape == (2, 1500)

sciope/utilities/priors/prior_base.py

@@ -35,7 +35,7 @@ def __init__(self, name, use_logger=False):
         self.use_logger = use_logger
 
     @abstractmethod
-    def pdf(self, x, log = False):
+    def pdf(self, x, log=False):
         """
         Evaluate the PDF of the sample
         :param x: the point or collection of points to evaluate the pdf at

sciope/utilities/priors/uniform_prior.py

@@ -34,6 +34,7 @@ def __init__(self, space_min, space_max, use_logger=False):
         Set up a uniform prior corresponding to the space bounded by:
         :param space_min: the lowerbound of each variable/dimension
         :param space_max: the upperbound of each variable/dimension
+        :param use_logger: whether logging is enabled or disabled
         """
         self.name = 'Uniform'
         self.lb = space_min

setup.py

@@ -1,6 +1,6 @@
 """A setuptools based setup module for sciope.
 See:
-http://hellanderlab.research.it.uu.se
+https://github.com/sciope/sciope
 """
 
 # Always prefer setuptools over distutils
@@ -27,7 +27,7 @@
     long_description=long_description,
 
     # The project's main homepage.
-    url='https://hellanderlab.research.it.uu.se',
+    url='https://github.com/sciope/sciope',
 
     # Author details
     author='Prashant Singh; Fredrik Wrede; Andreas Hellander',
@@ -76,7 +76,7 @@
         'numpy',
         'scipy',
         'scikit-learn',
-        'tsfresh',
+        'tsfresh==0.15.0',
         'ipywidgets',
         'dask',
         'distributed',
@@ -92,9 +92,6 @@
         'test': ['coverage'],
     },
 
-    dependency_links=['https://github.com/GPflow/GPflow/archive/0.5.0.tar.gz#egg=GPflow-0.5.0',
-                      'https://github.com/GPflow/GPflowOpt/archive/v0.1.0.tar.gz#egg=GPflowOpt-0.1.0'],
-
     # If there are data files included in your packages that need to be
     # installed, specify them here.  If using Python 2.6 or less, then these
     # have to be included in MANIFEST.in as well.