Merge pull request #22 from fhswf/oa/streams/ca

OA-Streams: Introduction of Cluster's Size

src/mlpro_int_river/wrappers/clusteranalyzers/basics.py

@@ -29,10 +29,11 @@
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separation of particular algorithms into separate modules
 ## -- 2024-05-24  1.5.0     DA       Alignment with MLPro 2
+## -- 2024-05-25  1.5.1     SY       Introduction of size as a property, Refactoring
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.5.0 (2024-05-24)
+Ver. 1.5.1 (2024-05-25)
 
 This module provides wrapper root classes from River to MLPro, specifically for cluster analyzers. 
 
@@ -88,7 +89,7 @@ class WrClusterAnalyzerRiver2MLPro (WrapperRiver, ClusterAnalyzer):
 
     C_TYPE                  = 'River Cluster Analyzer'
 
-    C_CLUSTER_PROPERTIES    = [ cprop_centroid ]
+    C_CLUSTER_PROPERTIES    = [ cprop_centroid, cprop_size ]
 
 ## -------------------------------------------------------------------------------------------------
     def __init__( self,
@@ -145,13 +146,13 @@ def _adapt(self, p_inst_new : Instance) -> bool:
 
         # update MLPro clusters from river
         self.get_clusters()
-        self._update_clusters()
+        self._update_clusters(input_data)
 
         return True
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """

src/mlpro_int_river/wrappers/clusteranalyzers/clustream.py

@@ -28,10 +28,12 @@
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separated to own module
+## -- 2024-05-25  1.4.1     SY       Introduction of size as a property
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.4.0 (2024-05-07)
+Ver. 1.4.1 (2024-05-25)
+
 
 This module provides a wrapper class for the CluStream algorithm provided by River.
 
@@ -154,14 +156,23 @@ def __init__(self,
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """
 
+        updated_cls = self._river_algo.predict_one(input_data)
+
         for x in self._river_algo.centers.keys():
             related_cluster = self._clusters[x]
             related_cluster.centroid.value = list(self._river_algo.centers[x].values())
+
+            if x == updated_cls:
+                act_size = related_cluster.size.value
+                if act_size is not None:
+                    related_cluster.size.value = act_size+1
+                else:
+                    related_cluster.size.value = 1
 
 
 ## -------------------------------------------------------------------------------------------------
@@ -180,7 +191,7 @@ def get_clusters(self) -> dict[Cluster]:
             try:
                 related_cluster = self._clusters[x]
             except:
-                related_cluster = self._cls_cluster(p_id=x, p_visualize=self.get_visualization())   
+                related_cluster = self._cls_cluster(p_id=x, p_properties=self.C_CLUSTER_PROPERTIES, p_visualize=self.get_visualization())   
 
                 if self.get_visualization():
                     related_cluster.init_plot(p_figure = self._figure, p_plot_settings=self._plot_settings)             

src/mlpro_int_river/wrappers/clusteranalyzers/dbstream.py

@@ -28,10 +28,11 @@
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separated to own module
+## -- 2024-05-25  1.4.1     SY       Introduction of size as a property
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.4.0 (2024-05-07)
+Ver. 1.4.1 (2024-05-25)
 
 This module provides a wrapper class for the DBStream algorithm provided by River.
 
@@ -105,8 +106,6 @@ class WrRiverDBStream2MLPro (WrClusterAnalyzerRiver2MLPro):
 
     C_TYPE                  = 'River Cluster Analyzer DBSTREAM'
 
-    C_CLUSTER_PROPERTIES    = [ cprop_centroid ]
-
 ## -------------------------------------------------------------------------------------------------
     def __init__(self,
                  p_name:str = None,
@@ -138,14 +137,23 @@ def __init__(self,
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """
 
-        for _, (key, val) in enumerate(self._river_algo.micro_clusters.items()):
+        updated_cls = self._river_algo.predict_one(input_data)
+
+        for x, (key, val) in enumerate(self._river_algo.micro_clusters.items()):
             related_cluster = self._clusters[id(val)]
             related_cluster.centroid.value = list(self._river_algo.centers[key].values())
+
+            if x == updated_cls:
+                act_size = related_cluster.size.value
+                if act_size is not None:
+                    related_cluster.size.value = act_size+1
+                else:
+                    related_cluster.size.value = 1
 
 
 ## -------------------------------------------------------------------------------------------------

src/mlpro_int_river/wrappers/clusteranalyzers/denstream.py

@@ -28,10 +28,11 @@
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separated to own module
+## -- 2024-05-25  1.4.1     SY       Introduction of size as a property
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.4.0 (2024-05-07)
+Ver. 1.4.1 (2024-05-25)
 
 This module provides a wrapper class for the DenStream algorithm provided by River.
 
@@ -142,21 +143,29 @@ def __init__(self,
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """
 
         if self._river_algo.n_clusters != 0: 
+            updated_cls = self._river_algo.predict_one(input_data)
+
             for val in self._river_algo.p_micro_clusters.values():
                 related_cluster = self._clusters[id(val)]
 
                 list_center = []
-                for _, (_, val_center) in enumerate(val.x.items()):
+                for x, (_, val_center) in enumerate(val.x.items()):
                     list_center.append(val_center) 
 
                 try:
-                    related_cluster.centroid.value = list_center
+                    related_cluster.centroid.value = list_center        
+                    if x == updated_cls:
+                        act_size = related_cluster.size.value
+                        if act_size is not None:
+                            related_cluster.size.value = act_size+1
+                        else:
+                            related_cluster.size.value = 1
                 except:
                     pass
 

src/mlpro_int_river/wrappers/clusteranalyzers/kmeans.py

@@ -28,10 +28,11 @@
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separated to own module
+## -- 2024-05-25  1.4.1     SY       Introduction of size as a property
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.4.0 (2024-05-07)
+Ver. 1.4.1 (2024-05-25)
 
 This module provides a wrapper class for the KMeans algorithm provided by River.
 
@@ -133,16 +134,26 @@ def __init__(self,
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """
+
+        updated_cls = self._river_algo.predict_one(input_data)        
+
         for x in self._river_algo.centers.keys():
             related_cluster = self._clusters[x]
             list_center = []
             for y in range(len(self._river_algo.centers[x])):
                 list_center.append(self._river_algo.centers[x][y+1])
             related_cluster.centroid.value = list_center
+
+            if x == updated_cls:
+                act_size = related_cluster.size.value
+                if act_size is not None:
+                    related_cluster.size.value = act_size+1
+                else:
+                    related_cluster.size.value = 1
 
 
 ## -------------------------------------------------------------------------------------------------

src/mlpro_int_river/wrappers/clusteranalyzers/streamkmeans.py

@@ -28,10 +28,11 @@
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
 ## -- 2024-05-07  1.4.0     DA       Separated to own module
+## -- 2024-05-25  1.4.1     SY       Introduction of size as a property
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.4.0 (2024-05-07)
+Ver. 1.4.1 (2024-05-25)
 
 This module provides a wrapper class for the STREAMKMeans algorithm provided by River.
 
@@ -140,16 +141,25 @@ def __init__(self,
 
 
 ## -------------------------------------------------------------------------------------------------
-    def _update_clusters(self):
+    def _update_clusters(self, input_data):
         """
         This method is to update the centroids of each introduced cluster.
         """
+
+        updated_cls = self._river_algo.predict_one(input_data)
+
         for x in self._river_algo.centers.keys():
             try:
                 related_cluster = self._clusters[x]
                 list_center = []
                 for y in range(len(self._river_algo.centers[x])):
                     list_center.append(self._river_algo.centers[x][y+1])
+                if x == updated_cls:
+                    act_size = related_cluster.size.value
+                    if act_size is not None:
+                        related_cluster.size.value = act_size+1
+                    else:
+                        related_cluster.size.value = 1
                 try:
                     related_cluster.centroid.value = list_center
                 except:

test/howtos/oa/howto_oa_ca_001_run_kmeans_2d_static.py

@@ -10,10 +10,11 @@
 ## -- 2023-12-28  1.1.0     DA       Exchange of benchmark stream and number of clouds
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
 ## -- 2024-05-05  1.3.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.3.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.3.0 (2024-05-05)
+Ver. 1.3.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of static 2D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -129,6 +130,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 

test/howtos/oa/howto_oa_ca_002_run_kmeans_2d_dynamic.py

@@ -10,10 +10,11 @@
 ## -- 2023-12-28  1.1.0     DA       Exchange of benchmark stream and number of clouds
 ## -- 2024-01-05  1.1.1     SY       Replace algorithm to StreamKMeans
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.2.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.2.0 (2024-04-30)
+Ver. 1.2.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of dynamic 2D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -131,6 +132,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 

test/howtos/oa/howto_oa_ca_003_run_kmeans_2d_static_normalized.py

@@ -14,10 +14,11 @@
 ## -- 2023-12-21  1.1.1     SY       Refactoring
 ## -- 2024-02-02  1.2.0     SY       Parameters Optimization
 ## -- 2024-04-30  1.3.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.3.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.3.0 (2024-04-30)
+Ver. 1.3.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of normalized static 2D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -158,6 +159,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 

test/howtos/oa/howto_oa_ca_004_run_kmeans_2d_dynamic_normalized.py

@@ -12,10 +12,11 @@
 ## --                                - Add window to the workflow
 ## -- 2024-02-02  1.1.0     SY       Parameters Optimization
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.2.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.2.0 (2024-04-30)
+Ver. 1.2.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of normalized dynamic 2D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -165,6 +166,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 

test/howtos/oa/howto_oa_ca_005_run_kmeans_3d_static.py

@@ -10,10 +10,11 @@
 ## -- 2023-12-28  1.1.0     DA       Exchange of benchmark stream and number of clouds
 ## -- 2023-12-29  1.2.0     DA       Adjustments on cloud sizes and weights
 ## -- 2024-04-30  1.3.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.3.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.3.0 (2024-04-30)
+Ver. 1.3.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of static 3D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -132,6 +133,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 

test/howtos/oa/howto_oa_ca_006_run_kmeans_3d_dynamic.py

@@ -10,10 +10,11 @@
 ## -- 2023-12-28  1.1.0     DA       Exchange of benchmark stream and number of clouds
 ## -- 2024-01-05  1.1.1     SY       Replace algorithm to StreamKMeans
 ## -- 2024-04-30  1.2.0     DA       Alignment with MLPro 2
+## -- 2024-05-27  1.2.1     SY       Printing clusters' sizes
 ## -------------------------------------------------------------------------------------------------
 
 """
-Ver. 1.2.0 (2024-04-30)
+Ver. 1.2.1 (2024-05-27)
 
 This module demonstrates online cluster analysis of dynamic 3D random point clouds using the wrapped
 River implementation of stream algorithm KMeans. To this regard, the systematics of sub-framework 
@@ -132,6 +133,7 @@ def _setup(self, p_mode, p_ada: bool, p_visualize: bool, p_logging):
 myscenario.log(Log.C_LOG_TYPE_I, 'Number of clusters: ', number_of_clusters)
 for x in range(number_of_clusters):
     myscenario.log(Log.C_LOG_TYPE_I, 'Center of Cluster ', str(x+1), ': ', list(clusters[x].centroid.value))
+    myscenario.log(Log.C_LOG_TYPE_I, 'Size of Cluster ', str(x+1), ': ', clusters[x].size.value)
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')
 myscenario.log(Log.C_LOG_TYPE_I, '-------------------------------------------------------')