code: Notes for Copenhagen data

code/parse_copenhagen.py

@@ -16,6 +16,64 @@ import matplotlib.pyplot as plt
 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
+'''
 
 # https://cloud.delkappa.com/s/ACMsDr2jnW3b6Np
 # Copenhagen data format
@@ -111,6 +169,12 @@ 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)) | (cont_data[:, uid_b] == float(usr))]
+  #   print(int(usr), len(usr_cont))
+  # exit()
   '''
     Get contacts for user 14
     9378 contacts
@@ -126,8 +190,6 @@ def main(args):
     usr_cont_cur = np.copy(usr_cont)
     # Remove goal landmarks from contacts
     for c in cont:
-      # print(usr_cont)
-      # exit()
       usr_cont_cur = usr_cont_cur[(usr_cont_cur[:, uid_a] != c) & (usr_cont_cur[:, uid_b] != c)]
     # Check
     print(g, (len(usr_cont) - len(usr_cont_cur))/len(usr_cont))