parse_copenhagen: Working only with ego

code/parse_copenhagen.py

@@ -17,62 +17,62 @@ import time
 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
-705 5
-76 4875
-77 25
-79 8892
-81 8944
-88 17254
+  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
+  705 5
+  76 4875
+  77 25
+  79 8892
+  81 8944
+  88 17254
 '''
 
 # https://cloud.delkappa.com/s/ACMsDr2jnW3b6Np
@@ -123,7 +123,7 @@ def main(args):
     Get users' landmarks
   '''
   # Get all users
-  usrs = np.unique(np.concatenate((cont_data[:, uid_a], cont_data[:, uid_b]), 0))
+  usrs = np.unique(cont_data[:, uid_a])
   # Check each user
   goal = [.2, .4, .6, .8]
   # Get users' data from previous parsing
@@ -135,7 +135,7 @@ def main(args):
       print('Checking %d (%d%%: %d/%d)... ' %(usr, (usr_i + 1)*100/len(usrs), usr_i + 1, len(usrs)), end='', flush=True)
       usrs_expt_cur = []
       # User's contacts
-      usr_cont = cont_data[(cont_data[:, uid_a] == usr) | (cont_data[:, uid_b] == usr)]
+      usr_cont = cont_data[cont_data[:, uid_a] == usr]
       # For each goal
       for_expt = True
       for g in goal:
@@ -152,7 +152,7 @@ def main(args):
             # 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[:, uid_b] != pos_c)]
+              usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_a] != 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
@@ -167,12 +167,14 @@ def main(args):
               print('[NOK]')
               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
   # usrs = np.unique(usrs_expt[:, 0])
   # for usr in usrs:
-  #   usr_cont = cont_data[(cont_data[:, uid_a] == float(usr)) | (cont_data[:, uid_b] == float(usr))]
+  #   usr_cont = cont_data[cont_data[:, uid_a] == float(usr)]
   #   print(int(usr), len(usr_cont))
   # exit()
   '''
@@ -181,7 +183,7 @@ def main(args):
   '''
   usr = '14'
   # All user contacts
-  usr_cont = cont_data[(cont_data[:, uid_a] == float(usr)) | (cont_data[:, uid_b] == float(usr))]
+  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:
@@ -190,7 +192,7 @@ 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[:, uid_b] != c)]
+      usr_cont_cur = usr_cont_cur[usr_cont_cur[:, uid_a] != c]
     # Check
     print(g, (len(usr_cont) - len(usr_cont_cur))/len(usr_cont))