clustering with agi dataset

src/data-hub/community-detection-microservice/app/configs/swagger.yml

@@ -72,7 +72,7 @@ paths:
                 schema:
                   $ref: "#/definitions/Cluster"
 
-  /cluster.png:
+  /cluster/cluster.png:
     get:
         operationId: "rest.cluster.get_image"
         tags:
@@ -86,6 +86,32 @@ paths:
                 description: "Successful operation"
   
   
+  /agi/cluster:
+    get:
+        operationId: "rest.agi_cluster.get"
+        tags:
+            - "Clusters"
+        summary: "Get clustered data"
+        parameters: []
+        responses:
+            200:
+                description: "Successful operation"
+                schema:
+                  $ref: "#/definitions/Cluster"
+
+  /agi/cluster/cluster.png:
+    get:
+        operationId: "rest.agi_cluster.get_image"
+        tags:
+            - "Clusters"
+        summary: "Get clustered data as image"
+        parameters: []
+        produces:
+            - "image/png"
+        responses:
+            200:
+                description: "Successful operation"
+              
 definitions:
   Location:
     type: "object"

src/data-hub/community-detection-microservice/app/db/agi/agi_repository.py

+
+import json
+from typing import List, Dict
+
+class AgiRepository:
+    def getLocations(self) -> List:
+        locations = []
+
+        travels = self.readDataFromFile()
+
+        # only take started travels
+        travels = [t for t in travels if t['status'] >= 2]
+        for travel in travels:
+
+            locations.append(self.location(travel["id"], travel['startPlace.latitude'], travel['startPlace.longitude'], 0, ''))
+            continue # todo work on locations
+
+            # todo number of complete travels with startlocation and user data
+            num_complete_travels = min(len(travel['startedBy']), len(travel['users']))
+            for i in range(num_complete_travels):
+
+                cur_location = travel['startedBy'][i]
+                cur_user = travel['users'][i]
+                locations.append(
+                    self.location(f'{travel["id"]}-{cur_location["moment"]}',
+                            cur_location['coordinate']['latitude'],
+                            cur_location['coordinate']['longitude'],
+                            cur_location['moment'],
+                            # todo user in travel startedBy not available from dataset - currently using user list 
+                            cur_user['userId']
+                ))
+
+        return locations
+
+    def readDataFromFile(self) -> List[Dict]:
+        with open('./db/agi/travels.json', 'r') as f_travels:
+            travels = json.loads(f_travels.read())
+
+        return travels
+
+    def location(self, id_, lat, long_, timestamp, username) -> dict:
+        return {
+            "id": id_,
+            'latitude': lat,
+            'longitude': long_,
+            "timestamp": timestamp,
+            "username": username
+        }
+

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

@@ -17,4 +17,4 @@ def api_root():
 
 # start app
 if __name__ == '__main__':    
-    app.run(host='0.0.0.0', port=5000, debug=True, use_reloader=True) 
+    app.run(host='0.0.0.0', port=5000, debug=True) 

src/data-hub/community-detection-microservice/app/rest/agi_cluster.py

+import io
+from flask import request, Response
+from db.agi.agi_repository import AgiRepository
+from processing.clusterer import Clusterer
+from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
+
+repo = AgiRepository()
+clusterer = Clusterer()
+
+def get():
+    locations = repo.getLocations()
+
+    clusters = clusterer.run(locations)
+
+    return clusters
+
+def get_image():
+    locations = repo.getLocations()
+
+    fig = clusterer.draw_locations(locations)
+
+    output = io.BytesIO()
+    FigureCanvas(fig).print_png(output)
+    return Response(output.getvalue(), mimetype="image/png")
\ No newline at end of file