code: Checking for duplicates

code/parse_copenhagen.py

@@ -78,8 +78,10 @@ def main(args):
             print('[Error: %s]' %(e))
     except Exception as e:
       print('[Error: %s]' %(e))
+    # Remove duplicates
+    cont_data = np.unique(np.array(cont, np.float32), axis=0)
     # Save to results
-    lmdk_lib.save_data(args, np.array(cont, np.float32), 'cont')
+    lmdk_lib.save_data(args, cont_data, 'cont')
   '''
     Get users' landmarks
   '''
@@ -129,30 +131,30 @@ def main(args):
               break
     # Save to results
     lmdk_lib.save_data(args, np.array(usrs_expt, str), 'usrs_expt')
-  # # Get all users
-  # usrs = np.unique(usrs_expt[:, 0])
-  # for usr in usrs:
-  #   usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
-  #   print(int(usr), len(usr_cont))
-  # exit()
-  '''
-    Get contacts for user 623
-    9378 contacts
-  '''
-  usr = '623'
-  # All user contacts
-  usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
-  # All user landmarks for different goals
-  usr_lmdk = usrs_expt[usrs_expt[:, 0] == usr]
-  for g in goal:
-    # Get goal landmarks
-    cont = ast.literal_eval(usr_lmdk[usr_lmdk[:, 1] == str(g)][0][2])
-    usr_cont_cur = np.copy(usr_cont)
-    # Remove goal landmarks from contacts
-    for c in cont:
-      usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_b] != c]
-    # Check
-    print(g, (len(usr_cont) - len(usr_cont_cur))/len(usr_cont))
+  # Get all users
+  usrs = np.unique(usrs_expt[:, 0])
+  for usr in usrs:
+    usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
+    print(int(usr), len(usr_cont))
+  exit()
+  # '''
+  #   Get contacts for user 623
+  #   9378 contacts
+  # '''
+  # usr = '623'
+  # # All user contacts
+  # usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
+  # # All user landmarks for different goals
+  # usr_lmdk = usrs_expt[usrs_expt[:, 0] == usr]
+  # for g in goal:
+  #   # Get goal landmarks
+  #   cont = ast.literal_eval(usr_lmdk[usr_lmdk[:, 1] == str(g)][0][2])
+  #   usr_cont_cur = np.copy(usr_cont)
+  #   # Remove goal landmarks from contacts
+  #   for c in cont:
+  #     usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_b] != c]
+  #   # Check
+  #   print(g, (len(usr_cont) - len(usr_cont_cur))/len(usr_cont))
 
 
 '''