parse_copenhagen: Better results

code/parse_copenhagen.py

@@ -18,61 +18,22 @@ import zipfile
 
 '''
   usr n
-  109 6549
-  112 3277
-  113 4202
-  139 8681
-  14 9378
-  145 4192
-  147 4428
-  157 11503
-  164 8045
-  17 4238
-  176 19732
-  178 7738
-  186 3289
-  190 13843
-  216 6126
-  236 3810
-  244 10563
-  262 17567
-  264 9792
-  274 11572
-  275 4316
-  287 13394
-  299 7373
-  304 6502
-  317 3892
-  324 11349
-  344 14479
-  374 5512
-  383 18751
-  389 3438
-  396 7004
-  453 3957
-  474 7692
-  480 4009
-  504 2622
-  505 11200
-  561 4304
-  570 2136
-  576 6553
-  58 12755
-  593 3429
-  595 4982
-  619 6189
-  633 5229
-  657 10941
-  658 1658
-  663 11461
-  688 19794
-  691 6755
+  311 8193
+  317 2044
+  323 1642
+  366 3406
+  368 5099
+  478 2616
+  486 3112
+  508 4981
+  585 9443
+  595 4459
+  616 4355
+  623 1586
+  637 4479
+  664 4315
+  688 19728
   705 5
-  76 4875
-  77 25
-  79 8892
-  81 8944
-  88 17254
 '''
 
 # https://cloud.delkappa.com/s/ACMsDr2jnW3b6Np
@@ -140,19 +101,18 @@ def main(args):
       for_expt = True
       for g in goal:
         if for_expt:
+          # User's contacts
+          conts = np.unique(usr_cont[:, uid_b])
           # Possible contacts
           pos_cont = []
-          usrs_cur = list(usrs)
-          # Remove user
-          usrs_cur.remove(usr)
           # Check for every possible contact
-          for u in usrs_cur:
+          for c in conts:
             # Add possible contacts gradually
-            pos_cont.append(int(u))
+            pos_cont.append(int(c))
             # Remove from user contacts
             usr_cont_cur = np.copy(usr_cont)
             for pos_c in pos_cont:
-              usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_a] != pos_c]
+              usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_b] != pos_c]
             # Compare the difference
             diff = (len(usr_cont) - len(usr_cont_cur))/len(usr_cont)
             # Check if it is close enough to what we need
@@ -164,24 +124,22 @@ def main(args):
                 usrs_expt += usrs_expt_cur
               break
             elif diff > g:
-              print('[NOK]')
+              print('[%.2f]' %(diff))
               for_expt = False
               break
-      # This user is no bueno
-      print('[NOK]')    
     # Save to results
     lmdk_lib.save_data(args, np.array(usrs_expt, str), 'usrs_expt')
-  # Get all users
+  # # 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 14
+    Get contacts for user 623
     9378 contacts
   '''
-  usr = '14'
+  usr = '623'
   # All user contacts
   usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
   # All user landmarks for different goals
@@ -192,7 +150,7 @@ def main(args):
     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_a] != c]
+      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))