-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_mav.py
1422 lines (1202 loc) · 53.3 KB
/
train_mav.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
import os
import sys
import json
from collections import defaultdict
import math
import io
import scipy
import collections
from scipy.sparse import csr_matrix
import numpy as np
try:
import psycopg2
except:
sys.path.insert(0, '/home/02608/grantdel/pythonlib/lib/python2.7/site-packages')
import psycopg2
from sklearn import linear_model
try:
os.environ['OBSPATH']
except:
print "ERROR: OBSPATH environment variable is not set"
sys.exit()
try:
os.environ['DB_CONN']
except:
#Looks like "dbname=topodb user=postgres host='localhost' port='5433' password='grant'"
print "ERROR: DB_CONN string is not set"
sys.exit()
demonyms = set([u'equatorial guineans', u'french southern territories', u'chinese', u'swedish',
u'malaysians', u'anguillans', u's\xe3o tom\xe9an', u'dutchwomen', u'turkish', u'incas', u'chadian',
u'iraqi', u'martiniquaises', u'futunans', u'georgians', u'tanzanian', u'burkinab\xe9',
u'burkinab\xe8', u'azerbaijanis', u'slovenians', u'bermudian', u'bahamians', u'omani',
u'nigerians', u'saudi', u'azeris', u'mozambicans', u'croatian', u'uruguayan', u'tunisian',
u'japanese', u'czechs', u'central african', u'letts', u'somalis', u'south africans', u'indian',
u'poles', u'faroese', u'aruban', u'omanis', u'sahraouis', u'belongers', u'laotians',
u'netherlanders', u'r\xe9unionnais', u'french', u'barbudan', u'republic of',
u'\xe5land islanders', u'malawians', u'canaleros', u'canadian', u'vietnamese', u'vanuatuan',
u'haitians', u'comorans', u'abkhazian', u'chapines', u'panamanian', u'ethiopian',
u'yemenis', u'south ossetian', u'italians', u'congolese', u'nepalese', u'hongkongers', u'gibraltarians',
u'syrians', u'belizeans', u'bissau-guineans', u'palestinian', u'spaniards', u'indians', u'samoans',
u'miquelonnais', u'australian', u'dominicans', u'belgian', u'northern marianans', u'bahamian', u'bulgarian',
u'brazilians', u'luxembourgers', u'georgian', u'costa ricans', u'luxembourgish', u'israelis', u'albanians',
u'none', u'icelanders', u'filipino', u'bouvet island', u'democratic republic of the', u'eritreans', u'costa rican',
u'mauritanians', u'sahrawis', u'bolivians', u'guadeloupe', u'hong kong', u'papua new guineans', u'tajiks', u'nigerien',
u'burundian', u'botswanan', u'wallis and futuna islanders', u'guamanian', u'kuwaiti', u'montenegrin', u'macanese',
u'egyptian', u'slovenian', u'mongolians', u'mongols', u'magyars', u'kyrgyz', u'nauruan', u'south sudanese', u'lao',
u'lithuanian', u'fijian', u'maltese', u'kenyans', u'hellenes', u'equatorial guinean', u'guaran\xedes', u'malians',
u'moroccans', u'turks and caicos island', u'tokelauans', u'greek', u'burmese', u'kirgiz', u'mexican', u'futunan',
u"democratic people's republic of", u'maldivians', u'guanacos', u'palauans', u'austrian', u'argentines', u'gabonese',
u'emirian', u'kittitian', u'argentine', u'guatemalan', u'koreans', u'monacans', u'antarctic residents', u'beninois',
u'ugandans', u'chileans', u'saint-martinoise', u'iranians', u'maldivian', u'northern irish', u'mauritanian', u'algerian',
u'new zealanders', u'pakistanis', u'irishwomen', u'surinamese', u'european', u'namibian', u'guatemalans', u'trinis',
u'botswanans', u'slovaks', u'nigerian', u'christmas island', u'honduran', u'andorrans', u'kirghiz', u'caymanian', u'tanzanians',
u'afghans', u'greeks', u'uzbekistani', u'mcdonald islands', u'sahraouian', u'barbudans', u'dutch', u'britons', u'kiwis',
u'malagasy', u'french guianese', u'mon\xe9gasque', u'abkhaz', u'timorese', u'egyptians', u'bhutanese', u'northern marianan',
u'qataris', u'vincentians', u'french polynesian', u'norwegian', u'british', u's\xe3o tom\xe9ans', u'somalian', u'barbadian',
u'american', u'belgians', u'sierra leonean', u"people's republic of", u'cubans', u'christmas islanders', u'arubans', u'ayisyen',
u'micronesian', u'svalbard', u'gambians', u'americans', u'spanish', u'antarctic', u'tongan', u'cambodians', u'r\xe9unionnaises',
u'palauan', u'cabo verdean', u'basotho (singular mosotho)', u'mauritians', u'federated states of', u'mon\xe9gasques', u'sint eustatius',
u'gibraltar', u'danish', u'solomon islanders', u'herzegovinian', u'burundians', u'greenlandic', u'sri lankans', u'saudi arabians', u'eritrean',
u'saint helenian', u'thai', u'moldovan', u'ghanaian', u'pakistani', u'barundi', u'colombians', u'iranian', u'latvians', u'saudi arabian',
u'saint-martinois', u'norfolk island', u'north korean', u'belarusians', u'catrachos', u'tunisians', u'motswana', u'philippine', u'scotswomen',
u'ecuadorian', u'angolans', u'mongolian', u'abkhazians', u'trinibagonians', u'tobagonian', u'scotsmen', u'salvadoran', u'russians', u'samoan',
u'llanitos', u'south korean', u'zimbabwean', u'equatoguineans', u'barth\xe9lemoises', u'mauritian', u'sri lankan', u'british virgin island',
u'emirians', u'bissau-guinean', u'ukrainians', u'statian', u'qatari', u'sint maarten', u'germans', u'estonian', u'cuban', u'slovene',
u'cura\xe7aoans', u'papuans', u'banyarwanda', u'liechtenstein', u'greenlanders', u'sint maartener', u'moldovans', u'see taiwan',
u'hong kongese', u'pitcairn islanders', u'malian', u'german', u'austrians', u'moroccan', u'jordanian', u'cambodian', u'puerto ricans',
u'senegalese', u'singaporean', u'dutchmen', u'mozambican', u'romanian', u'saints', u'filipinos', u'bruneian', u'namibians', u'anguillan',
u'caymanians', u'montserratians', u'nevisians', u'hungarian', u'comorian', u'emiri', u'tajikistani', u'malawian', u'niueans', u'nevisian',
u'turks and caicos islanders', u'emiratis', u'tuvaluans', u'malaysian', u'\xe5land island', u'comoran', u'angolan', u'uzbeks', u'scots',
u'papuan', u'kazakhs', u'philippinean', u'channel islander', u'cook island', u'tuvaluan', u'american samoan', u'nicas', u'taiwanese',
u'cura\xe7aoan', u'welshmen', u'guambat', u'martiniquais', u'swedes', u'europeans', u'indonesians', u'israeli', u'turks', u'guyanese',
u'northern irishmen', u'mexicans', u'salvadorans', u'puerto rican', u'welshwomen', u'colombian', u'niuean', u'belizean', u'persians',
u'fijians', u'bolivian', u'sahrawian', u'liberian', u'armenian', u'slovenes', u'bulgarians', u'filipinas', u'new caledonian', u'manx',
u'cypriots', u'bahraini', u'montenegrins', u'martinican', u'canadians', u'algerians', u'libyan', u'jamaicans', u'i-kiribati',
u'scottish', u'togolese', u'swazi', u'barth\xe9lemois', u'saudis', u'seychellois', u'rwandans', u'bamar', u'american samoans',
u'laos', u'guadeloupians', u'saba', u'swazis', u'miquelonnaises', u'cabo verdeans', u'new caledonians', u'charr\xfaas', u'somali',
u'grenadians', u'emirati', u'zambian', u'venezuelans', u'sammarinese', u'kyrgyzstani', u'croatians', u'lebanese', u'zambians',
u'frenchwomen', u'bruneians', u'saint vincentian', u'see other words for british', u'uzbek', u'herzegovinians', u'ivorians',
u'zimbabweans', u'kosovan', u'ivorian', u'cook islanders', u'hondurans', u'nicaraguans', u'danes', u'bahrainis', u'saint-pierraises',
u'saint vincentians', u'sudanese', u'kuwaitis', u'kosovar', u'cariocas', u'bonaire dutch', u'saint-pierrais', u'norfolk islander', u'magyar',
u'paraguayans', u'comorians', u'azerbaijani', u'french polynesians', u'tajikistanis', u'ghanaians', u'northern irishwomen',
u'south sandwich islands', u'singapore', u'welsh', u'frenchmen', u'djiboutians', u'guineans', u'sahrawi', u'afghan', u'kazakh',
u'montserratian', u'mahoran', u'saint lucian', u'central africans', u'iraqis', u'basotho', u'liberians', u'english', u'swiss',
u'micronesians', u'the', u'kazakhstanis', u'andorran', u'motswana (sing. batswana)', u'bermudans', u'liechtensteiners', u'belarusian',
u'macedonian', u'bangladeshis', u'falkland islanders', u'cocos islanders', u'polish', u'cameroonian', u'icelandic', u'serbs',
u'saba dutch', u'ugandan', u'solomon island', u'ethiopians', u'r\xe9unionese', u'falkland island', u'ascension and tristan da cunha',
u'wallisian', u'channel island', u'indonesian', u'kyrgyzstanis', u'antiguans', u'vincentian', u'ni-vanuatu', u'tongans', u'finnish',
u'bosnians', u'u.s. virgin island', u'serbian', u'bermudan', u'pinoys', u'italian', u'portuguese', u'bonaire', u'chadians',
u'quisqueyanos', u'czech', u'finns', u'barbadians', u'kittitians', u'laotian', u'sierra leoneans', u'bajans', u'new zealand',
u'uruguayans', u'marshallese', u'libyans', u'papua new guinean', u'jan mayen', u'ukrainian', u'saint helenians', u'luxembourg',
u'malinese', u'jordanians', u'englishwomen', u'british virgin islanders', u'tobagonians', u'trinidadian', u'englishmen',
u'hungarians', u'paraguayan', u'romanians', u'saint-martinoises', u'kazakhstani', u'chilean', u'vatican citizens', u'gambian',
u'kosovars', u'hellenic', u'albanian', u'hongkies', u'maubere', u'antiguan', u'latvian', u'peruvians', u'palestinians',
u'trinidadians', u'boricuas', u'united states', u'slovak', u'u.s. virgin islanders', u'beninese', u'pitcairn island',
u'cocos island', u'serbians', u'grenadian', u'bangladeshi', u'biot', u'brazilian', u'venezuelan', u'estonians',
u'cameroonians', u'azeri', u'south georgia', u'bosnian', u'bermudians', u'hongers', u'south ossetians', u'turkmen',
u'russian', u'haitian', u'wallisians', u'statians', u'dominican', u'macedonians', u'somalians', u'us', u'below',
u'persian', u'jamaican', u'uk', u'vatican', u'saint lucians', u'beninoises', u'irish', u'cypriot', u'peruvian',
u'guamanians', u'singaporeans', u'yemeni', u'croats', u'syrian', u'surinamers', u'south african', u'nicaraguan',
u'guinean', u'ticos', u'djiboutian', u'armenians', u'seychelloises', u'nz', u'turkmens', u'nigeriens', u'rwandan',
u'netherlandic', u'kenyan', u'irishmen', u'tokelauan', u'uzbekistanis', u'aussies', u'equatoguinean', u'monacan', u'nauruans',
u'wallis and futuna', u'nepali', u'australians', u'ecuadorians', u'heard island', u'norwegians', u'mahorans',
u'lithuanians', u'panamanians'])
def flatten(l):
for el in l:
if isinstance(el, collections.Iterable) and not isinstance(el, basestring):
for sub in flatten(el):
yield sub
else:
yield el
def getPossibleTopoRegions(cur, topo, cntry_alt, region_alt, state_alt, pplc_alt, country_tbl="countries_2012", region_tbl="regions_2012", state_tbl="states_2012", geonames_tbl="geonames_all"):
region_entry = []
norm_topo = topo.title()
names = tuple([norm_topo, topo])
cntry_gid_list = list()
cntry_gid_list.extend(flatten([cntry_alt.get(g) for g in names if g in cntry_alt]))
cntry_gid_list.extend([-99])
region_gid_list = list()
region_gid_list.extend(flatten([region_alt.get(g) for g in names if g in region_alt]))
region_gid_list.extend([-99])
state_gid_list = list()
state_gid_list.extend(flatten([state_alt.get(g) for g in names if g in state_alt]))
state_gid_list.extend([-99])
pplc_gid_list = list()
pplc_gid_list.extend(flatten(pplc_alt.get(g, -99) for g in names))
SQL1 = "SELECT p2.region_name FROM %s as p1, customgrid as p2 WHERE p1.gid IN %s or p1.postal IN %s or p1.abbrev IN %s or p1.name_long IN %s and ST_DWithin(p1.geog, p2.geog, 10.0);" % (country_tbl, '%s', '%s', '%s', '%s')
SQL2 = "SELECT p2.region_name FROM %s as p1, customgrid as p2 WHERE p1.gid IN %s and ST_DWithin(p1.geog, p2.geog, 10.0);" % (region_tbl, '%s')
SQL3 = "SELECT p2.region_name FROM %s as p1, customgrid as p2 WHERE p1.gid IN %s or p1.abbrev in %s or p1.postal in %s and ST_DWithin(p1.geog, p2.geog, 10.0);" % (state_tbl, '%s', '%s', '%s')
SQL4 = "SELECT p2.region_name FROM %s as p1, customgrid as p2 WHERE p1.gid IN %s and ST_DWithin(p1.geog, p2.geog, 10.0);" % (geonames_tbl, '%s')
#print "Got here"
#print SQL1
#print "Countries"
cur.execute(SQL1, (tuple(cntry_gid_list), names, names, names))
returns = cur.fetchall()
for row in returns:
region_entry.append(row[0])
#print "!!!Found Gazet Match!!!"
#print region_entry[-1]
#print "States"
cur.execute(SQL2, (tuple(region_gid_list), ))
returns = cur.fetchall()
for row in returns:
region_entry.append(row[0])
#print "!!!Found Gazet Match!!!"
#print region_entry[-1]
#print "Regions"
cur.execute(SQL3, (tuple(state_gid_list), names, names))
returns = cur.fetchall()
for row in returns:
region_entry.append(row[0])
#print "!!!Found Gazet Match!!!"
#print region_entry[-1]
#print "PPL"
cur.execute(SQL4, (tuple(pplc_gid_list), ))
returns = cur.fetchall()
for row in returns:
region_entry.append(row[0])
#print "!!!Found Gazet Match!!!"
#print region_entry[-1]
#print topo
#print set(region_entry)
return set(region_entry)
def getAltGazets(country_tbl="countries_2012", region_tbl="regions_2012", state_tbl="states_2012", geonames_tbl="geonames_all"):
#These queries are designed to pull all the alternate names from the geonames, country, state, and region tables. Alternate names are used in later steps to enhance gazetteer matching
SQL1 = "SELECT p1.gid, p1.name, p1.name_long, p1.geonames_gid, p1.altnames FROM %s as p1 ;" % country_tbl
SQL2 = "SELECT p1.gid, p1.name, p1.name_long, p1.geonames_gid, p1.altnames FROM %s as p1 ;" % region_tbl
SQL3 = "SELECT p1.gid, p1.name, p1.geonames_gid, p1.altnames FROM %s as p1 ;" % state_tbl
SQL4 = "SELECT p1.gid, p1.name, p1.asciiname, p1.alternames FROM %s as p1 where p1.featurecode = 'PPLC' or p1.featurecode = 'PPLA' or p1.featurecode = 'PPLA2' or p1.featurecode = 'PPL';" % geonames_tbl
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
print "Creating Alt Gazets"
cur.execute(SQL1)
cntry_alt = {}
for row in cur.fetchall():
alist = [row[1], row[2]]
if row[4] is not None:
alist.extend(row[4].split(','))
#print alist
for w in alist:
cntry_alt.setdefault(w, set()).add(row[0])
#cntry_alt.setdefault(row[0], list()).append(alist)
cur.execute(SQL3)
state_alt = {}
for row in cur.fetchall():
alist = [row[1], row[2]]
if row[3] is not None:
alist.extend(row[3].split(','))
#print alist
for w in alist:
state_alt.setdefault(w, set()).add(row[0])
#state_alt.setdefault(row[0], list()).append(alist)
cur.execute(SQL2)
region_alt = {}
for row in cur.fetchall():
alist = [row[1], row[2]]
#print row
if len(row) > 3 and row[4] is not None:
alist.extend(row[4].split(','))
#print alist
for w in alist:
region_alt.setdefault(w, set()).add(row[0])
#region_alt.setdefault(row[0], list()).append(alist)
cur.execute(SQL4)
pplc_alt = {}
for row in cur.fetchall():
alist = [row[1], row[2]]
#print row
if len(row) > 3 and row[3] is not None:
alist.extend(row[3].split(','))
for w in alist:
pplc_alt.setdefault(w, set()).add(row[0])
print "Done Creating Alt Names"
print "Length of PPL AltNames: ", len(pplc_alt)
conn.close()
return cntry_alt, region_alt, state_alt, pplc_alt
class transition_model_discrim:
trans_data = []
feature_index = {}
label_index = {}
custom_regions = []
country_names = []
region_bin_dict = {}
def __init__(self):
self.trans_counts = {}
def load(self, direct):
import ParseLGL
conn = psycopg2.connect(os.environ['DB_CONN'])
self.load_custom_regions()
self.load_country_names()
self.load_region_bin_dict()
cur = conn.cursor()
m = 0
for xml_infile in os.listdir(direct):
print xml_infile
m += 1
wordref, toporef, domain = ParseLGL.parse_xml(os.path.join(direct, xml_infile))
self.trans_data = featurize_transition_discrim(wordref, toporef, domain, cur, self.country_names)
conn.close()
def train(self):
X = []
Y = []
feature_index, label_index = {}, {}
i, j, y = 0, 0, 0
row_indexes, col_indexes, values, labels = [], [], [], []
feature_index[""] = 0
i += 1
for line in self.trans_data:
if len(line) > 0:
label = line[0]
feats = line[1]
row_indexes.append(len(values))
# Add the intercept constant
col_indexes.append(0)
values.append(1.0)
j += 1
if label not in label_index:
label_index[label] = y
y += 1
labels.append(label_index[label])
for f in feats:
if f not in feature_index:
feature_index[f] = i
i += 1
col_indexes.append(feature_index[f])
values.append(1.0)
row_indexes.append(row_indexes[-1] + 1)
X = csr_matrix((scipy.array(values, dtype=scipy.float64), scipy.array(col_indexes),
scipy.array(row_indexes)), shape=(j, i), dtype=scipy.float64)
Y = scipy.array(labels)
model = linear_model.LogisticRegression(penalty='l2', fit_intercept=False, C=10.0, solver='newton-cg')
model.fit(X, Y)
coefs = model.coef_
print label_index
print feature_index
print coefs.tolist()
self.label_index = label_index
self.feature_index = feature_index
self.weights = coefs.tolist()
#feature_set should be output of discrim_featurize function
def log_prob_dict(self, feature_set):
feature_vector = np.zeros(len(self.feature_index))
prob_dict = {}
if len(self.label_index) > 2:
for label in self.label_index:
label_sum = 0.0
label_sum += self.weights[self.label_index[label]][0]
for feat in feature_set:
if feat in self.feature_index:
label_sum += self.weights[self.label_index[label]][self.feature_index[feat]]
prob_dict[label] = math.log(math.exp(label_sum) / (1.0 + math.exp(label_sum)))
return prob_dict
elif len(self.label_index) == 2:
discount = .10
inv_index = {v: k for k, v in self.label_index.items()}
label = inv_index[1]
label_sum = 0.0
label_sum += self.weights[0][0]
for feat in feature_set:
if feat in self.feature_index:
label_sum += self.weights[0][self.feature_index[feat]]
prob_dict[label] = math.exp(label_sum) / (1.0 + math.exp(label_sum))
label2 = inv_index[0]
prob_dict[label2] = (1.0 - (math.exp(label_sum) / (1.0 + math.exp(label_sum))))
if "CONTINENT/GLOBAL" not in self.label_index:
prob_dict2 = {}
prob_dict2["CONTINENT/GLOBAL"] = math.log(sum([prob_dict[l]*discount for l in prob_dict]))
for pb in prob_dict:
prob_dict2[pb] = math.log(prob_dict[pb] - (prob_dict[pb]*discount))
#print "HAVING TO BACK OFF TO CONTINENT/GLOBAL INTERP"
#prob_dict2[pb]
#print feature_set
#sys.exit()
return prob_dict2
if 'LOCAL/ADJACENT' not in self.label_index:
prob_dict2 = {}
prob_dict2['LOCAL/ADJACENT'] = math.log(sum([prob_dict[l]*discount for l in prob_dict]))
for pb in prob_dict:
prob_dict2[pb] = math.log(prob_dict[pb] - (prob_dict[pb]*discount))
#print "HAVING TO BACK OFF TO CONTINENT/GLOBAL INTERP"
#prob_dict2[pb]
#print feature_set
#sys.exit()
return prob_dict2
return prob_dict
def load_region_bin_dict(self):
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
region_bin_dict = {}
print "Loading Region Bin Dict"
for region in self.custom_regions:
SQL = "SELECT p1.region_name, p2.region_name, ST_DWithin(p1.geog, p2.geog, 161000.0) from customgrid as p1, customgrid as p2 where p1.region_name = %s;"
cur.execute(SQL, (region, ))
results = cur.fetchall()
results.append([region, region, True])
for row in results:
#print row
reg1 = row[0]
reg2 = row[1]
if reg1 == reg2:
if reg1 not in region_bin_dict:
region_bin_dict[reg1] = {}
region_bin_dict[reg1][reg2] = 'SAME'
elif row[2] == True:
if reg1 not in region_bin_dict:
region_bin_dict[reg1] = {}
region_bin_dict[reg1][reg2] = "LOCAL/ADJACENT"
else:
if reg1 not in region_bin_dict:
region_bin_dict[reg1] = {}
region_bin_dict[reg1][reg2] = "CONTINENT/GLOBAL"
self.region_bin_dict = region_bin_dict
conn.close()
def load_country_names(self):
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
country_names = []
SQL = "SELECT p1.name, p1.postal, p1.abbrev, p1.name_long, p1.altnames from countries_2012 as p1;"
cur.execute(SQL)
results = cur.fetchall()
for row in results:
for name in row:
if name != None:
if ',' in name:
for nm in name.split(','):
country_names.append(unicode(nm.decode('utf-8').lower()))
else:
country_names.append(unicode(name.decode('utf-8').lower()))
#for n in country_names:
# print n
self.country_names = set(country_names)
conn.close()
def load_custom_regions(self):
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
SQL = "SELECT p1.region_name from customgrid as p1;"
cur.execute(SQL)
returns = cur.fetchall()
for name in returns:
self.custom_regions.append(name[0])
conn.close()
def discrim_featurize(prev_toponame, cur_toponame, token_dist, country_names):
obs_features = []
Token_Bins = {'adjacent':[0, 4], 'sentence':[5, 25], 'paragraph':[26, 150], 'document':[151, 4000]}
#Add the duplicate toponym feature
if prev_toponame.lower() == cur_toponame.lower():
obs_features.append('SAME_TOPO')
#Add the token distance bin feature
#tokebin = get_tokenbin(Token_Bins, token_dist)
#if tokebin in Token_Bins:
# obs_features.append(tokebin)
#'County' is in the toponym
if 'county' in cur_toponame.lower():
obs_features.append('CUR_COUNTY')
if 'county' in prev_toponame.lower():
obs_features.append('PREV_COUNTY')
#Add the demonym features
if len(isDemonym(prev_toponame, demonyms)) > 0:
obs_features.append('PREV_DEMONYM')
if len(isDemonym(cur_toponame, demonyms)) > 0:
obs_features.append('CUR_DEMONYM')
#Add the is a country name feature
if len(isCountryName(prev_toponame, country_names)) > 0:
obs_features.append('PREV_COUNTRYNAME')
if len(isCountryName(cur_toponame, country_names)) > 0:
obs_features.append('CUR_COUNTRYNAME')
return obs_features
class transition_model:
trans_counts = {}
custom_regions = []
def __init__(self):
self.trans_counts = {}
#Takes 'classic_xml variety'
#direct=os.environ['TRANSITION_DIR']
def load(self, direct):
import ParseLGL
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
m = 0
for xml_infile in os.listdir(direct):
print xml_infile
m += 1
wordref, toporef, domain = ParseLGL.parse_xml(os.path.join(direct, xml_infile))
self.trans_counts = featurize_transition_gen(wordref, toporef, domain, cur)
#print self.trans_counts
#print toporef
#print len(wordref)
#print self.trans_counts
conn.close()
self.load_custom_regions()
def load_custom_regions(self):
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
SQL = "SELECT p1.region_name from customgrid as p1;"
cur.execute(SQL)
returns = cur.fetchall()
for name in returns:
self.custom_regions.append(name[0])
conn.close()
def binomial_prob_dict(self):
p_dict = {}
for k in sorted(self.trans_counts.keys()):
p_dict[k] = {}
for r in self.custom_regions:
p_dict[k][r] = self.binomial_prob(k, r)
return p_dict
#Additive smoothing
def binomial_prob(self, prev_region, current_region, additive_smoothing=3.0):
if prev_region not in self.trans_counts:
cur_region_n = additive_smoothing
else:
cur_region_n = self.trans_counts[prev_region].get(current_region, 0.0) + additive_smoothing
prev_total = 0.0
for r in self.custom_regions:
if prev_region in self.trans_counts:
prev_total += (self.trans_counts[prev_region].get(r, 0.0) + additive_smoothing)
else:
prev_total += additive_smoothing
prob = cur_region_n / float(prev_total)
return prob
class lang_model:
obs_counts = {}
custom_regions = []
def __init__(self):
self.obs_counts = {}
#load obscounts dict given obs directory
def load(self, direct=os.environ['OBSPATH']):
j = 0
m = 0
self.obs_counts['global'] = defaultdict(float)
for f in os.listdir(direct):
print "Loading", f
print m, "/", len(os.listdir(direct))
m += 1
if m < 300:
fp = os.path.join(direct, f)
geocat = f.split('_uni_bigram')[0].replace('_', ' ')
self.obs_counts[geocat] = {}
uni_total = 0.0
bi_types = 0
bi_total = 0.0
uni_types = 0
with open(fp, 'rb') as r:
rdict = json.loads(r.read())
for k in rdict:
j += 1
if j % 200000 == 0:
print j
self.obs_counts[geocat][k] = rdict[k]
self.obs_counts['global'][k] += rdict[k]
if '|' not in k:
uni_total += rdict[k]
uni_types += 1
else:
bi_types += 1
bi_total += rdict[k]
if rdict[k] < 0.0:
print "Shit is fucked up"
print geocat, k, rdict[k]
sys.exit()
#second_word = k.split('|')[1]
#first_word = k.split('|')[0]
#self.obs_counts[geocat]['$SECOND_WORD$|'+second_word] = self.obs_counts[geocat].get('$SECOND_WORD$|'+second_word, 0) + rdict[k]
#self.obs_counts[geocat]['$FIRST_WORD$|'+first_word] = self.obs_counts[geocat].get('$FIRST_WORD$|'+first_word, 0) + rdict[k]
self.obs_counts[geocat]['$UNI_TOTAL$'] = uni_total
self.obs_counts[geocat]['$UNI_TYPES$'] = uni_types
self.obs_counts[geocat]['$BI_TYPES$'] = bi_types
self.obs_counts[geocat]['$BI_TOTAL$'] = bi_total
self.load_custom_regions()
def load_custom_regions(self):
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
SQL = "SELECT p1.region_name from customgrid as p1;"
cur.execute(SQL)
returns = cur.fetchall()
for name in returns:
self.custom_regions.append(name[0])
#generate probability given unigram
def unigram_prob(smoothing="kneser-ney"):
pass
#generate probability of word in region assuming each geog has independent distribution
def bigram_prob_indep(self, bigram, smoothing="simple-interp", lamb=.5):
probdict = {}
if smoothing=='simple-interp':
firstword = bigram.split('|')[0]
secondword = bigram.split('|')[1]
for geocat in self.obs_counts:
if geocat != 'global':
#Add in some absolute discounting
bi_discount = self.obs_counts[geocat]['$BI_TOTAL$'] / self.obs_counts[geocat]['$BI_TYPES$']
uni_discount = self.obs_counts[geocat]['$UNI_TOTAL$'] / self.obs_counts[geocat]['$UNI_TYPES$']
uni_prob_first = (max(self.obs_counts[geocat].get(firstword, 0.0) - uni_discount, 0.0) / self.obs_counts[geocat]['$UNI_TOTAL$'])
uni_prob_second = (max(self.obs_counts[geocat].get(secondword, 0.0) - uni_discount, 0.0) / self.obs_counts[geocat]['$UNI_TOTAL$'])
#bi_discount = 2.0
bi_prob = max(self.obs_counts[geocat].get(bigram, 0.0)-bi_discount, 0.0) / self.obs_counts[geocat].get(firstword, 1.0)
#print geocat, bi_prob, uni_prob_first, uni_prob_second
#print "bigram-count: ", self.obs_counts[geocat].get(bigram, 0.0), "unigram-first:", self.obs_counts[geocat].get(firstword, 0.0), "unigram-total:", self.obs_counts[geocat]['$UNI_TOTAL$']
interp_prob = lamb * bi_prob + (((1.0 - lamb)/2.0) * uni_prob_first) + (((1.0 - lamb)/2.0) * uni_prob_second)
probdict[geocat] = interp_prob
return probdict
#generate probability given bigram. assumes all geographies are in same distribution
def bigram_prob(self, bigram, smoothing="simple-interp-laplace", lamb=.5):
probdict = {}
if smoothing=='kneser-ney':
firstword = bigram.split('|')[0]
secondword = bigram.split('|')[1]
'''for geocat in self.obs_counts:
discount = self.obs_counts[geocat]['$BI_TOTAL$'] / self.obs_counts[geocat]['$BI_TYPES$']
normalizing_constant = (discount / self.obs_counts[geocat].get(firstword, 1.0)) * float(self.obs_counts[geocat]['$FIRST_WORD$|'+firstword])
right_term = normalizing_constant * (float(self.obs_counts[geocat].get('$SECOND_WORD$|'+secondword, 0.0)) / float(self.obs_counts[geocat]['$BI_TYPES$']))
left_term = max(self.obs_counts[geocat].get(bigram, 0.0)-discount, 0.0) / self.obs_counts[geocat].get(firstword, 1.0)
print geocat, left_term, right_term, normalizing_constant, discount
probdict[geocat] = right_term + left_term'''
elif smoothing=='simple-interp':
firstword = bigram.split('|')[0]
secondword = bigram.split('|')[1]
for geocat in self.obs_counts:
if geocat != 'global':
uni_prob_first = self.obs_counts[geocat].get(firstword, 0.0) / self.obs_counts['global'].get(firstword, 1.0)
uni_prob_second = self.obs_counts[geocat].get(secondword, 0.0) / self.obs_counts['global'].get(secondword, 1.0)
bi_prob = self.obs_counts[geocat].get(bigram, 0.0) / self.obs_counts['global'].get(bigram, 1.0)
interp_prob = lamb * bi_prob + (((1.0 - lamb)/2.0) * uni_prob_first) + (((1.0 - lamb)/2.0) * uni_prob_second)
probdict[geocat] = interp_prob
elif smoothing=="simple-interp-laplace":
firstword = bigram.split('|')[0]
secondword = bigram.split('|')[1]
for geocat in self.custom_regions:
if geocat != 'global':
uni_prob_first = (self.obs_counts[geocat].get(firstword, 0.0) + 1.0) / float(self.obs_counts['global'].get(firstword, 0.0) + (len(self.custom_regions)-1.0))
uni_prob_second = (self.obs_counts[geocat].get(secondword, 0.0) + 1.0) / float(self.obs_counts['global'].get(secondword, 0.0) + (len(self.custom_regions)-1.0))
bi_prob = (self.obs_counts[geocat].get(bigram, 0.0) + 1.0) / float(self.obs_counts['global'].get(bigram, 0.0) + (len(self.custom_regions)-1.0))
interp_prob = (lamb * bi_prob) + (((1.0 - lamb)/2.0) * uni_prob_first) + (((1.0 - lamb)/2.0) * uni_prob_second)
probdict[geocat] = interp_prob
if interp_prob < 0.0 or interp_prob > 1.0:
print "Shit is fucked up"
print bigram
print "bigram: ", bi_prob
print "interp: ", interp_prob
numerator = (self.obs_counts[geocat].get(bigram, 0.0) + 1.0)
denom = float(self.obs_counts['global'].get(bigram, 0.0) + (len(self.custom_regions)-1.0))
denom_part1 = self.obs_counts['global'].get(bigram, 0.0)
denom_part2 = (len(self.custom_regions)-1.0)
print numerator
print denom
print denom_part1
print denom_part2
sys.exit()
'''for geocat in self.obs_counts:
#Add in some absolute discounting
bi_discount = self.obs_counts[geocat]['$BI_TOTAL$'] / self.obs_counts[geocat]['$BI_TYPES$']
uni_discount = self.obs_counts[geocat]['$UNI_TOTAL$'] / self.obs_counts[geocat]['$UNI_TYPES$']
print "============="
print "bi_discount", bi_discount
print "uni_discount", uni_discount
uni_prob_first = (max(self.obs_counts[geocat].get(firstword, 0.0) - uni_discount, 0.0) / self.obs_counts[geocat]['$UNI_TOTAL$'])
uni_prob_second = (max(self.obs_counts[geocat].get(secondword, 0.0) - uni_discount, 0.0) / self.obs_counts[geocat]['$UNI_TOTAL$'])
#bi_discount = 2.0
bi_prob = max(self.obs_counts[geocat].get(bigram, 0.0)-bi_discount, 0.0) / self.obs_counts[geocat].get(firstword, 0.0)
print geocat, bi_prob, uni_prob_first, uni_prob_second
print "bigram-count: ", self.obs_counts[geocat].get(bigram, 0.0), "unigram-first:", self.obs_counts[geocat].get(firstword, 0.0), "unigram-total:", self.obs_counts[geocat]['$UNI_TOTAL$']
interp_prob = lamb * bi_prob + (((1.0 - lamb)/2.0) * uni_prob_first) + (((1.0 - lamb)/2.0) * uni_prob_second)
probdict[geocat] = interp_prob'''
return probdict
def get_distbin(Dist_Bins, dist_transition):
for b in Dist_Bins:
if dist_transition <= Dist_Bins[b][1] and dist_transition >= Dist_Bins[b][0]:
return b
return 'global'
def get_tokenbin(Dist_Bins, dist_transition):
for b in Dist_Bins:
if dist_transition <= Dist_Bins[b][1] and dist_transition >= Dist_Bins[b][0]:
return b
return 'document'
def getRegionBin(current_region, prev_region, cur):
#print current_region
#print prev_region
if current_region == prev_region:
return 'SAME'
SQL_DIST = "SELECT ST_DWithin(p1.geog, p2.geog, 161000.0) from customgrid as p1, customgrid as p2 where p1.region_name = %s and p2.region_name = %s;"
cur.execute(SQL_DIST, (current_region, prev_region))
results = cur.fetchall()
#print results
if results[0][0] == True:
return "LOCAL/ADJACENT"
#SQL_DIST = "SELECT ST_DWithin(p1.geog, p2.geog, 1500000.0) from customgrid as p1, customgrid as p2 where p1.region_name = %s and p2.region_name = %s;"
#cur.execute(SQL_DIST, (current_region, prev_region))
#results = cur.fetchall()
#if results[0][0] == True:
# return "COUNTRY"
return "CONTINENT/GLOBAL"
#RETURN COUNTRY if toponym is a name or alt name of a country
def isCountryName(toponym, country_names):
if toponym.lower() in country_names:
return 'COUNTRY'
else:
return ''
#RETURN Demonym is toponym is a demonym
def isDemonym(toponym, demonyms):
if toponym.lower() in demonyms:
return 'Demonym'
else:
return ''
#Get region given latitutde, longitude, DB cur
def getRegion(lat, lon, cur):
SQL_REGION = "SELECT p2.region_name, ST_Distance(ST_GeographyFromText('SRID=4326;POINT(%s %s)'), p2.geog)/1000.0 from customgrid as p2;" % (lon, lat)
cur.execute(SQL_REGION)
results = cur.fetchall()
results.sort(key=lambda x: x[1])
#print results
return results[0][0]
def featurize_transition_discrim(wordref, toporef, domain, cur, country_names):
j = 0
#Dist_Bins = {'local':[0.0, 161.0], 'region':[161.1, 500.0], 'country':[500.1, 1500.0], 'global':[1501.1, 15000.0]}
prev_region = '#START#'
transition_data = []
for i in sorted(toporef.keys()):
j += 1
#print i, toporef[i]
lon = toporef[i][1]['long']
lat = toporef[i][1]['lat']
toponym = toporef[i][0]
docid = toporef[i][-2]
wid = i
regions = []
#print domain, results[0][0]
current_region = getRegion(lat, lon, cur)
if j > 1:
prev_lon = last_topo[1]['long']
prev_lat = last_topo[1]['lat']
prev_wid = last_topo[-1]
prev_docid = last_topo[-2]
prev_toponame = last_topo[0]
#print last_topo[0], "->", toporef[i][0]
token_dist = i - last_topo[-1]
label = getRegionBin(current_region, prev_region, cur)
obs_features = discrim_featurize(prev_toponame, toponym, token_dist, country_names)
transition_data.append([label, obs_features])
#print transition_data[-1]
last_topo = toporef[i]
last_topo[-1] = i
if j > 0:
prev_region = current_region
return transition_data
def featurize_transition_gen(wordref, toporef, domain, cur, transition_dict):
j = 0
Dist_Bins = {'local':[0.0, 161.0], 'region':[161.1, 500.0], 'country':[500.1, 1500.0], 'global':[1501.1, 15000.0]}
Token_Bins = {'adjacent':[0, 4], 'sentence':[5, 25], 'paragraph':[26, 150], 'document':[151, 4000]}
prev_region = '#START#'
for i in sorted(toporef.keys()):
j += 1
#print i, toporef[i]
lon = toporef[i][1]['long']
lat = toporef[i][1]['lat']
toponym = toporef[i][0]
docid = toporef[i][-2]
wid = i
regions = []
#print domain, results[0][0]
if j > 1:
prev_lon = last_topo[1]['long']
prev_lat = last_topo[1]['lat']
prev_wid = last_topo[-1]
prev_docid = last_topo[-2]
#print last_topo[0], "->", toporef[i][0]
current_region = getRegion(lat, lon, cur)
if prev_region not in transition_dict:
transition_dict[prev_region] = {}
transition_dict[prev_region][current_region] = (transition_dict[prev_region].get(current_region, 0) + 1)
last_topo = toporef[i]
last_topo[-1] = i
if j > 1:
prev_region = current_region
current_region = "#END#"
if prev_region not in transition_dict:
transition_dict[prev_region] = {}
transition_dict[prev_region][current_region] = (transition_dict[prev_region].get(current_region, 0) + 1)
return transition_dict
#topo_context_dict[t] = {'entry':toporef[t], 'context':d}
def test_viterbi_poly(LM, TM, directory="/work/02608/grantdel/corpora/LGL/articles/dev_testsplit1", poly_table_name = "lgl_dev_classic"):
import ParseLGL
import viterbi
out_test = "test_output3.txt"
ot = io.open(out_test, 'w', encoding='utf-8')
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
cor = 0
total = 0
obs_sequence = []
for f in os.listdir(directory):
#print f
wordref, toporef, domain = ParseLGL.parse_xml(os.path.join(directory, f))
topo_context_dict = ParseLGL.getTopoContexts(wordref, toporef, window=1)
ordered_tkeys = sorted(topo_context_dict.keys())
obs = [topo_context_dict[topo]['context'].keys() for topo in ordered_tkeys]
#print "==="
#print "obs"
#print obs
#print "==="
states = TM.custom_regions
if len(obs) > 0:
prob, prob_path = viterbi.viterbi(obs, states, TM, LM)
zipped_preds = zip(prob_path, [toporef[topo] for topo in ordered_tkeys])
#print "prob path", zipped_preds
for pred in zipped_preds:
pred_region = pred[0]
lat = float(pred[1][1]['lat'])
lon = float(pred[1][1]['long'])
did = pred[1][-2]
wid = pred[1][-1]
#print "did: ", did
#print "wid: ", wid
SQL_ACC = "SELECT ST_DWithin(p1.polygeog2, p2.geog, 160000) from customgrid as p2, %s as p1 where p2.region_name = %s and p1.docid = %s and p1.wid = %s;" % (poly_table_name, '%s', '%s', '%s') #print SQL_ACC
cur.execute(SQL_ACC, (pred_region, did, wid))
returns = cur.fetchall()
#print returns
if returns[0][0] == None:
SQL_POINT = "SELECT ST_Distance(ST_GeographyFromText('SRID=4326;POINT(%s %s)'), p2.geog)/1000.0 from customgrid as p2 where p2.region_name = %s;" % (lon, lat, '%s')
#print SQL_ACC
cur.execute(SQL_POINT, (pred_region, ))
returns = cur.fetchall()
if returns[0][0] < 160.0:
cor += 1
#print "backed off to point acc and found CORRECT"
elif returns[0][0] == True:
cor += 1
total += 1
#print "viterbi poly total: ", total
try:
ot.write(unicode(pred_region) + u'|' + unicode(pred[1][0]) + u'|' + unicode(lat) + u'|' + unicode(lon) + u'|' + unicode(returns[0][0]))
ot.write(u'\n')
except:
print "=========="
print "error writing"
print pred
print "VITERBI ACC POLY:"
print cor, "/", total
print float(cor)/float(total)
ot.close()
conn.close()
def test_viterbi_discrim_poly(LM, TM, directory="/work/02608/grantdel/corpora/LGL/articles/dev_testsplit1", poly_table_name = "lgl_dev_classic"):
import ParseLGL
import viterbi
out_test = "test_output3.txt"
ot = io.open(out_test, 'w', encoding='utf-8')
conn = psycopg2.connect(os.environ['DB_CONN'])
cur = conn.cursor()
cor = 0
total = 0
for f in os.listdir(directory):
obs_sequence = []
#print f
wordref, toporef, domain = ParseLGL.parse_xml(os.path.join(directory, f))
topo_context_dict = ParseLGL.getTopoContexts(wordref, toporef, window=1)
ordered_tkeys = sorted(topo_context_dict.keys())
obs = [[topo, topo_context_dict[topo]['entry'][0], topo_context_dict[topo]['context'].keys()] for topo in ordered_tkeys]
#print obs
#print "==="
j = 0
for o in obs:
j += 1
topo = o[1]
topo_tokeid = o[0]
if j > 1:
toke_dist = topo_tokeid - prev_topo_tokeid
trans_features = discrim_featurize(prev_topo, topo, toke_dist, TM.country_names)
obs_sequence.append([o[2], trans_features])
else:
obs_sequence.append([o[2], []])
prev_topo = topo
prev_topo_tokeid = o[0]
#print "obs"
#print obs
#print "==="
states = TM.custom_regions
if len(obs_sequence) > 0:
prob, prob_path = viterbi.viterbi_discrim(obs_sequence, states, TM, LM, cur)
zipped_preds = zip(prob_path, [toporef[topo] for topo in ordered_tkeys])
print "prob path", zipped_preds
for pred in zipped_preds:
pred_region = pred[0]
lat = float(pred[1][1]['lat'])
lon = float(pred[1][1]['long'])
did = pred[1][-2]
wid = pred[1][-1]
#print "pred:", pred_region
#print "did: ", did
#print "wid: ", wid
SQL_ACC = "SELECT ST_DWithin(p1.polygeog2, p2.geog, 160000) from customgrid as p2, %s as p1 where p2.region_name = %s and p1.docid = %s and p1.wid = %s;" % (poly_table_name, '%s', '%s', '%s') #print SQL_ACC
cur.execute(SQL_ACC, (pred_region, did, wid))
returns = cur.fetchall()
#print returns
if returns[0][0] == None:
SQL_POINT = "SELECT ST_Distance(ST_GeographyFromText('SRID=4326;POINT(%s %s)'), p2.geog)/1000.0 from customgrid as p2 where p2.region_name = %s;" % (lon, lat, '%s')
#print SQL_ACC
cur.execute(SQL_POINT, (pred_region, ))
returns = cur.fetchall()
if returns[0][0] < 160.0:
cor += 1
#print "backed off to point acc and found CORRECT"
elif returns[0][0] == True:
cor += 1
total += 1
#print "viterbi poly total: ", total
try:
ot.write(unicode(pred_region) + u'|' + unicode(pred[1][0]) + u'|' + unicode(lat) + u'|' + unicode(lon) + u'|' + unicode(returns[0][0]))
ot.write(u'\n')
except:
print "=========="
print "error writing"
print pred
print "VITERBI DISCRIM POLY ACC:"
print cor, "/", total