Merge pull request #8 from UniStuttgart-VISUS/organized_imports

Organization of import statements

example.py

@@ -1,31 +1,31 @@
-import uadapy as ua
+from uadapy import distribution
+from uadapy.dr import uamds
 import uadapy.data as data
-import uadapy.dr.uamds as uamds
+from uadapy.data import load_iris_normal
 import uadapy.plotting.plots1D as plots1D
 import uadapy.plotting.plots2D as plots2D
 import uadapy.plotting.plotsND as plotsND
 import numpy as np
 import matplotlib.pyplot as plt
 
 
-
 def example_uamds():
     distribs_hi = data.load_iris_normal()
     plots1D.plot_1d_distribution(distribs_hi, num_samples=100, plot_types='boxplot')
-    distribs_lo = uamds.uamds(distribs_hi, dims=2)
+    distribs_lo = uamds(distribs_hi, dims=2)
     plots2D.plot_contour(distribs_lo)
     #plots2D.plot_contour_bands(distribs_lo, 10000, 128, None, [5, 25, 55, 75, 95])
     #plotsND.plot_contour(distribs_lo, 10000, 128, None, [5, 25, 50, 75, 95])
     #plotsND.plot_contour_samples(distribs_lo, 10000, 128, None, [5, 25, 50, 75, 95])
 
 def example_kde():
-    samples = np.random.randn(1000, 2)
-    distr = ua.distribution.distribution(samples)
+    samples = np.random.randn(1000,2)
+    distr = distribution(samples)
     plots2D.plot_contour(distr)
 
 def example_uamds_1d():
     distribs_hi = data.load_iris_normal()
-    distribs_lo = uamds.uamds(distribs_hi, dims=4)
+    distribs_lo = uamds(distribs_hi, dims=4)
     labels = ['setosa','versicolor','virginica']
     titles = ['sepal length','sepal width','petal length','petal width']
     colors = ['red','green', 'blue']

tests/api_consistency_test.py

@@ -0,0 +1,39 @@
+# This file is to test consistency of uadapy's API.
+# It calls most commonly used functions with default arguments and named arguments
+# as the corrsponding documentation suggests. If calling a function errors, it is
+# due to a change in the API or changes that result in breaking previously working
+# functionality. If an API change was intentional, the test calls in this file need
+# to be updated accordingly. In any case, this script should never error.
+
+# make script aware of parent directory where uadapy is located
+import os, sys
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+def test_data_module():
+    import uadapy.data
+    uadapy.data.load_iris()
+    uadapy.data.load_iris_normal()
+
+
+def test_dr_module():
+    import uadapy.dr
+    import uadapy.distribution
+    import numpy as np
+    # list of distributions (normal distributions estimated from random data
+    distribs = [uadapy.distribution(np.random.rand(10, 3), name='Normal') for _ in range(4)]
+    uadapy.dr.uapca(distributions=distribs, dims=2)
+    uadapy.dr.uamds(distributions=distribs, dims=2)
+
+
+def test_plotting_module():
+    # TODO
+    pass
+
+
+def main():
+    test_data_module()
+    test_dr_module()
+
+
+if __name__ == '__main__':
+    main()

uadapy/test_distrib.py → tests/test_distrib.py

@@ -1,9 +1,12 @@
-import traceback
+# make script aware of parent directory where uadapy is located
+import os, sys
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
+import traceback
+from uadapy import distribution
 import numpy as np
 import scipy as sp
 import scipy.stats as st
-import distribution as ds
 
 
 
@@ -68,11 +71,11 @@ def test_distrib_class():
 
     # initialize distribution object for each of the scipy distribs (univariate)
     for scipi_distrib in model_1D:
-        distrib = ds.distribution(scipi_distrib)
+        distrib = distribution(scipi_distrib)
     # initialize distribution object for each of the scipy distribs (multivariate)
     for scipi_distrib in model_nD:
         try:
-            distrib = ds.distribution(scipi_distrib)
+            distrib = distribution(scipi_distrib)
             cov = distrib.cov()
             if cov.shape[0] != n or cov.shape[1] != n:
                 raise RuntimeError(f"shape expected to be {n} x {n}, but was {cov.shape}")
@@ -83,9 +86,6 @@ def test_distrib_class():
             traceback.print_exception(e)
 
 
-
-
-
 if __name__ == '__main__':
     test_distrib_class()
 

uadapy/__init__.py

@@ -0,0 +1,3 @@
+from .distribution import distribution
+
+__all__ = ['distribution']

uadapy/data/__init__.py

@@ -0,0 +1 @@
+from .data import *

uadapy/data.py → uadapy/data/data.py

@@ -1,6 +1,6 @@
 from sklearn import datasets
 import numpy as np
-import uadapy.distribution as ua
+from uadapy import distribution
 
 def load_iris_normal():
     """
@@ -10,7 +10,7 @@ def load_iris_normal():
     iris = datasets.load_iris()
     dist = []
     for c in np.unique(iris.target):
-        dist.append(ua.distribution(np.array(iris.data[iris.target == c]), "Normal"))
+        dist.append(distribution(np.array(iris.data[iris.target == c]), "Normal"))
     return dist
 
 def load_iris():
@@ -21,5 +21,5 @@ def load_iris():
     iris = datasets.load_iris()
     dist = []
     for c in np.unique(iris.target):
-        dist.append(ua.distribution(np.array(iris.data[iris.target == c])))
+        dist.append(distribution(np.array(iris.data[iris.target == c])))
     return dist

uadapy/distribution.py

@@ -47,7 +47,7 @@ def pdf(self, x: np.ndarray | float) -> np.ndarray | float:
         if isinstance(self.model, np.ndarray):
             return self.kde.pdf(x.T)
         if not hasattr(self.model, 'pdf'):
-            raise AttributeError(f"The model has no pdf.{self.model.__class__.__name__}")
+            raise AttributeError(f"The model has no pdf. {self.model.__class__.__name__}")
         else:
             return self.model.pdf(x)
 
@@ -90,7 +90,7 @@ def skew(self) -> np.ndarray | float:
             return stats.skew(self.model)
         if hasattr(self.model, 'stats') and callable(self.model.stats):
             return self.model.stats(moments='s')
-        if isinstance(self.model, multivariate_t_frozen):
+        if isinstance(self.model, mv.multivariate_t_frozen):
             return 0
 
     def kurt(self) -> np.ndarray | float:

uadapy/dr/__init__.py

@@ -0,0 +1,2 @@
+from .uamds import *
+from .uapca import *

uadapy/dr/uamds.py

@@ -12,8 +12,8 @@
 import numpy as np
 from scipy.spatial import distance_matrix
 from scipy.optimize import minimize
-import uadapy as ua
 from scipy.stats import multivariate_normal
+from uadapy import distribution
 
 
 def precalculate_constants(normal_distr_spec: np.ndarray) -> tuple:
@@ -546,7 +546,7 @@ def uamds(distributions: list, dims: int=2):
         result = apply_uamds(means, covs, dims)
         distribs_lo = []
         for (m, c) in zip(result['means'], result['covs']):
-            distribs_lo.append(ua.distribution.distribution(multivariate_normal(m, c)))
+            distribs_lo.append(distribution(multivariate_normal(m, c)))
         return distribs_lo
     except Exception as e:
         raise Exception(f'Something went wrong. Did you input normal distributions? Exception:{e}')

uadapy/dr/uapca.py

@@ -1,5 +1,5 @@
 import numpy as np
-import uadapy as ua
+from uadapy import distribution
 from scipy.stats import multivariate_normal
 
 def uapca(distributions, dims: int):
@@ -18,7 +18,7 @@ def uapca(distributions, dims: int):
         means_pca, covs_pca = transform_uapca(means, covs, dims)
         dist_pca = []
         for (m, c) in zip(means_pca, covs_pca):
-            dist_pca.append(ua.distribution.distribution(multivariate_normal(m, c)))
+            dist_pca.append(distribution(multivariate_normal(m, c)))
         return dist_pca
     except Exception as e:
         raise Exception(f'Something went wrong. Did you input normal distributions? Exception:{e}')

uadapy/plotting/plots1D.py

@@ -1,5 +1,5 @@
 import numpy as np
-import uadapy.distribution as dist
+from uadapy import distribution
 import matplotlib.pyplot as plt
 from math import ceil, sqrt
 import glasbey as gb
@@ -67,7 +67,7 @@ def setup_plot(distributions, num_samples, seed, fig=None, axs=None, colors=None
 
     samples = []
 
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
 
     # Calculate the layout of subplots

uadapy/plotting/plots2D.py

@@ -1,23 +1,23 @@
 import matplotlib.pyplot as plt
 import numpy as np
-import uadapy.distribution as dist
+from uadapy import distribution
 from numpy import ma
 from matplotlib import ticker
 
-def plot_samples(distribution, num_samples, **kwargs):
+def plot_samples(distributions, num_samples, **kwargs):
     """
     Plot samples from the given distribution. If several distributions should be
     plotted together, an array can be passed to this function
-    :param distribution: Distributions to plot
+    :param distributions: Distributions to plot
     :param num_samples: Number of samples per distribution
     :param kwargs: Optional other arguments to pass:
         xlabel for label of x-axis
         ylabel for label of y-axis
     :return:
     """
-    if isinstance(distribution, dist.distribution):
-        distribution = [distribution]
-    for d in distribution:
+    if isinstance(distributions, distribution):
+        distributions = [distributions]
+    for d in distributions:
         samples = d.sample(num_samples)
         plt.scatter(x=samples[:,0], y=samples[:,1])
     if 'xlabel' in kwargs:
@@ -56,7 +56,7 @@ def plot_contour(distributions, resolution=128, ranges=None, quantiles:list=None
         If a quantile is not between 0 and 100 (exclusive), or if a quantile results in an index that is out of bounds.
     """
 
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
     contour_colors = generate_spectrum_colors(len(distributions))
 
@@ -136,7 +136,7 @@ def plot_contour_bands(distributions, num_samples, resolution=128, ranges=None,
         If a quantile is not between 0 and 100 (exclusive), or if a quantile results in an index that is out of bounds.
     """
 
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
 
     # Sequential and perceptually uniform colormaps

uadapy/plotting/plotsND.py

@@ -1,6 +1,6 @@
 import matplotlib.pyplot as plt
 import numpy as np
-import uadapy.distribution as dist
+from uadapy import distribution
 import uadapy.plotting.utils as utils
 
 def plot_samples(distributions, num_samples, **kwargs):
@@ -11,7 +11,7 @@ def plot_samples(distributions, num_samples, **kwargs):
     :param kwargs: Optional other arguments to pass:
     :return:
     """
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
     # Create matrix
     numvars = distributions[0].dim
@@ -78,7 +78,7 @@ def plot_contour(distributions, num_samples, resolution=128, ranges=None, quanti
         If the dimension of the distribution is less than 2.
     """
 
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
     contour_colors = utils.generate_spectrum_colors(len(distributions))
     # Create matrix
@@ -193,7 +193,7 @@ def plot_contour_samples(distributions, num_samples, resolution=128, ranges=None
         If the dimension of the distribution is less than 2.
     """
 
-    if isinstance(distributions, dist.distribution):
+    if isinstance(distributions, distribution):
         distributions = [distributions]
     contour_colors = utils.generate_spectrum_colors(len(distributions))
     # Create matrix