fABBA 1.2.2__tar.gz → 1.2.4__tar.gz

{fABBA-1.2.2 → fABBA-1.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fABBA
-Version: 1.2.2
+Version: 1.2.4
 Summary: An efficient aggregation method for the symbolic representation of temporal data
 Home-page: https://github.com/nla-group/fABBA
 Author: Xinye Chen, Stefan Güttel

{fABBA-1.2.2 → fABBA-1.2.4}/fABBA/__init__.py

@@ -10,7 +10,7 @@ except (ModuleNotFoundError, ValueError):
     from .separate.aggregation import aggregate
     
     
-__version__ = '1.2.2'
+__version__ = '1.2.4'
 from .load_datasets import load_images, loadData
 from .fabba import (image_compress, image_decompress, ABBAbase, ABBA,
                                       get_patches, patched_reconstruction, fABBA,

{fABBA-1.2.2 → fABBA-1.2.4}/fABBA/extmod/fabba_agg_cm_win.pyx

@@ -63,7 +63,9 @@ cpdef aggregate(double[:,:] data, str sorting, double tol=0.5):
     cdef unsigned int lab=0, num_group # , nr_dist=0
     cdef double[:] clustc # starting point coordinates
     cdef double dist
-    cdef long long[:] labels = np.full(len_ind, -1, dtype=int) 
+    
+    cdef np.ndarray[int, ndim=1] labels = np.full(len_ind, -1, dtype=int) 
+        
     cdef list splist = list() # list of starting points
     cdef Py_ssize_t i, ii, j
     
@@ -117,5 +119,5 @@ cpdef aggregate(double[:,:] data, str sorting, double tol=0.5):
         
         lab += 1
         
-    return np.asarray(labels), splist # , nr_dist
+    return labels, splist # , nr_dist
 

{fABBA-1.2.2 → fABBA-1.2.4}/fABBA/jabba/jabba.py

@@ -517,8 +517,23 @@ class JABBA(object):
         if series.dtype !=  'float64':
             series = series.astype('float64')
             
+        if isinstance(series, np.ndarray):
+            if len(series.shape) == 1:
+                uni_dim = True
+            else:
+                uni_dim = False
+                
+        elif isinstance(series, list):
+            if not isinstance(series[0], list):
+                uni_dim = True
+            else:
+                uni_dim = False
+        else:
+            raise ValueError('Please enter time series with correct shape.')
+            
         n_jobs = self.n_jobs_init(n_jobs)
-        if len(series.shape) == 1: 
+        
+        if uni_dim: 
             # Partition time series for parallelism (for n_jobs > 1 or = -1) if it is univarite
             self.return_series_univariate = True # means the series is univariate,
                                        # so the reconstruction can automatically 
@@ -530,6 +545,15 @@ class JABBA(object):
         else:
             self.return_series_univariate = False
             
+            if self.recap_shape is not None:
+                if not isinstance(series, np.ndarray):
+                    series = np.asarray(series)
+
+                    if series.shape != self.recap_shape:
+                        raise ValueError('Please enter the input with consistent dimensions.')
+
+                series = series.reshape(-1, int(np.prod(self.recap_shape[1:])))
+            
         string_sequences = list()
         start_set = list()
         if n_jobs != 1:

{fABBA-1.2.2 → fABBA-1.2.4}/fABBA.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fABBA
-Version: 1.2.2
+Version: 1.2.4
 Summary: An efficient aggregation method for the symbolic representation of temporal data
 Home-page: https://github.com/nla-group/fABBA
 Author: Xinye Chen, Stefan Güttel

{fABBA-1.2.2 → fABBA-1.2.4}/setup.py

@@ -26,7 +26,7 @@ class CustomBuildExtCommand(build_ext):
         
 setup_args = {'name':"fABBA",
         'packages':setuptools.find_packages(),
-        'version':"1.2.2",
+        'version':"1.2.4",
         'cmdclass': {'build_ext': CustomBuildExtCommand},
         'install_requires':["numpy>=1.3.0", "scipy>=0.7.0", 
                             "requests", "pandas",