clustering based on time

src/data-hub/community-detection-microservice/app/processing/clusterer.py

@@ -21,7 +21,7 @@ class Clusterer:
             partition_info = labels
         )
 
-    def _draw_locations(self, locations:np.ndarray=None, centroids:np.ndarray=None, partition_info=None) -> plt.Figure:
+    def _draw_locations(self, locations:np.ndarray=None, centroids:np.ndarray=None, partition_info:List=None) -> plt.Figure:
         fig = plt.Figure()
         axis = fig.add_subplot(1, 1, 1)
 
@@ -55,6 +55,9 @@ class Clusterer:
 
         return labels.tolist()
 
+    def extract_location_data(self, locations: List[dict]) -> np.ndarray:
+        return np.asarray([(float(l['latitude']), float(l['longitude'])) for l in locations])
+
     def label_locations(self, locations:List[Dict], labels:List) -> List:
         if locations is None or labels is None:
             return
@@ -65,7 +68,7 @@ class Clusterer:
         for i in range(len(locations)):
             locations[i]['cluster_label'] = labels[i]
 
-    def run(self, locations:List[Dict]) -> Dict[int, List[Dict]]:
+    def cluster_locations(self, locations:List[Dict]) -> Dict[int, List[Dict]]:
         '''Returns a dictionary with identified clusters and their locations copied from the input'''
         if locations is None or len(locations) == 0:
             # raise Exception("locations has to contain something")
@@ -80,5 +83,25 @@ class Clusterer:
         
         return clusters
 
-    def extract_location_data(self, locations: List[dict]) -> np.ndarray:
-        return np.asarray([(float(l['latitude']), float(l['longitude'])) for l in locations])
\ No newline at end of file
+    def cluster_times(self, times:List[Dict]) -> Dict[int, List[Dict]]:
+
+        # times.sort(key=lambda x: x['timestamp'])
+
+        times1 = np.asarray([(float(t['timestamp']), float(0)) for t in times])
+        # print(times)
+    
+        dbsc = DBSCAN(eps = self.epsilon, min_samples = self.min_points)
+        dbsc = dbsc.fit(times1)
+        labels = dbsc.labels_
+        print(labels)
+
+        self.label_locations(times, labels)
+        times = [t for t in times if t['cluster_label'] != -1]
+        print(times)
+
+        info = [l for l in labels if l != -1]
+        print(info)
+
+        times1 = np.asarray([(float(t['timestamp']), float(0)) for t in times])
+        fig = self._draw_locations(locations=times1, partition_info=info)
+        fig.savefig('img.png')
\ No newline at end of file

src/data-hub/community-detection-microservice/app/run_clustering.py

@@ -12,16 +12,15 @@ from processing.clusterer import Clusterer
 
 DEBUG = False
 
-# used to cluster the users based on their main location
-user_clusterer = Clusterer()
-
 repo = Repository()
 
 
 def run_location_clustering():
-    all_location_traces = repo.get_locations()
+    user_clusterer = Clusterer()
+
+    all_location_traces = repo.get_agi_locations()
 
-    cluster_result = user_clusterer.run(
+    cluster_result = user_clusterer.cluster_locations(
         [l.to_serializable_dict() for l in all_location_traces])
 
     clusters = [UserCluster(key, value)
@@ -30,6 +29,14 @@ def run_location_clustering():
     store_user_clusters(clusters)
 
 
+def run_time_clustering():
+    all_location_traces = repo.get_agi_locations()
+
+    user_clusterer = Clusterer(epsilon=10**5.8)
+
+    user_clusterer.cluster_times([l.to_serializable_dict() for l in all_location_traces])
+
+
 def store_user_clusters(user_clusters: List[UserCluster]):
     if DEBUG:
         print(user_clusters)
@@ -40,4 +47,4 @@ def store_user_clusters(user_clusters: List[UserCluster]):
 
 
 if __name__ == "__main__":
-    run_location_clustering()
+    run_time_clustering()