518:
92:
1312:
513:{\displaystyle {\hat {P}}(y\mid x_{1},\ldots x_{n})={\frac {\sum _{i:1\leq i\leq n\wedge F(x_{i})\geq m}{\hat {P}}(y,x_{i})\prod _{j=1}^{n}{\hat {P}}(x_{j}\mid y,x_{i})}{\sum _{y^{\prime }\in Y}\sum _{i:1\leq i\leq n\wedge F(x_{i})\geq m}{\hat {P}}(y^{\prime },x_{i})\prod _{j=1}^{n}{\hat {P}}(x_{j}\mid y^{\prime },x_{i})}}}
1171:
that makes the above independence assumption that is weaker (and hence potentially less harmful) than the naive Bayes' independence assumption. In consequence, each ODE should create a less biased estimator than naive Bayes. However, because the base probability estimates are each conditioned by two
1184:
whereby the classifier can be updated efficiently with information from new examples as they become available. It predicts class probabilities rather than simply predicting a single class, allowing the user to determine the confidence with which each classification can be made. Its probabilistic
1278:
is the number of classes. This makes it infeasible for application to high-dimensional data. However, within that limitation, it is linear with respect to the number of training examples and hence can efficiently process large numbers of training examples.
802:
1162:
1172:
variables rather than one, they are formed from less data (the training examples that satisfy both variables) and hence are likely to have more variance. AODE reduces this variance by averaging the estimates of all such ODEs.
983:
559:
1329:
837:
659:
1264:
1225:
617:
588:
1019:
845:
42:. It frequently develops substantially more accurate classifiers than naive Bayes at the cost of a modest increase in the amount of computation.
1376:
1348:
1180:
Like naive Bayes, AODE does not perform model selection and does not use tuneable parameters. As a result, it has low variance. It supports
1355:
1462:
1452:
1395:
1362:
623:
is a user specified minimum frequency with which a term must appear in order to be used in the outer summation. In recent practice
1344:
1333:
1457:
1322:
1300:
1369:
35:
1288:
1168:
39:
526:
1181:
797:{\displaystyle P(y\mid x_{1},\ldots x_{n})={\frac {P(y,x_{1},\ldots x_{n})}{P(x_{1},\ldots x_{n})}}.}
810:
1230:
1191:
593:
564:
1430:
38:
technique. It was developed to address the attribute-independence problem of the popular
1167:
This formula defines a special form of One
Dependence Estimator (ODE), a variant of the
1157:{\displaystyle P(y,x_{1},\ldots x_{n})=P(y,x_{i})\prod _{j=1}^{n}P(x_{j}\mid y,x_{i}).}
1446:
978:{\displaystyle P(y,x_{1},\ldots x_{n})=P(y,x_{i})P(x_{1},\ldots x_{n}\mid y,x_{i}).}
1311:
1434:
1422:
619:
is the frequency with which the argument appears in the sample data and
17:
1185:
model can directly handle situations where some data are missing.
1305:
486:
411:
322:
1423:"Not So Naive Bayes: Aggregating One-Dependence Estimators"
1291:
machine learning suite includes an implementation of AODE.
1233:
1194:
1022:
848:
813:
662:
596:
567:
529:
95:
50:
AODE seeks to estimate the probability of each class
1336:. Unsourced material may be challenged and removed.
1258:
1219:
1156:
977:
831:
796:
611:
582:
553:
512:
653:). By the definition of conditional probability
1421:Webb, G. I., J. Boughton, and Z. Wang (2005).
8:
1396:Learn how and when to remove this message
1247:
1232:
1208:
1193:
1142:
1123:
1107:
1096:
1083:
1055:
1039:
1021:
963:
944:
928:
909:
881:
865:
847:
812:
779:
763:
742:
726:
707:
695:
679:
661:
595:
566:
531:
530:
528:
498:
485:
472:
454:
453:
447:
436:
423:
410:
392:
391:
374:
339:
321:
316:
301:
282:
264:
263:
257:
246:
233:
209:
208:
191:
156:
149:
137:
121:
97:
96:
94:
1414:
1274:is the number of training examples and
7:
1345:"Averaged one-dependence estimators"
1334:adding citations to reliable sources
1188:AODE has computational complexity
554:{\displaystyle {\hat {P}}(\cdot )}
54:given a specified set of features
28:Averaged one-dependence estimators
25:
631:Derivation of the AODE classifier
86:). To do so it uses the formula
1310:
1321:needs additional citations for
1176:Features of the AODE classifier
1266:at classification time, where
1253:
1237:
1214:
1198:
1148:
1116:
1089:
1070:
1061:
1026:
969:
921:
915:
896:
887:
852:
785:
756:
748:
713:
701:
666:
606:
600:
577:
571:
548:
542:
536:
504:
465:
459:
429:
403:
397:
380:
367:
307:
275:
269:
239:
220:
214:
197:
184:
143:
108:
102:
1:
832:{\displaystyle 1\leq i\leq n}
1270:is the number of features,
1479:
1463:Statistical classification
1453:Classification algorithms
1435:10.1007/s10994-005-4258-6
1301:Cluster-weighted modeling
1259:{\displaystyle O(kn^{2})}
1220:{\displaystyle O(ln^{2})}
988:Under an assumption that
612:{\displaystyle F(\cdot )}
583:{\displaystyle P(\cdot )}
561:denotes an estimate of
36:classification learning
1260:
1221:
1169:naive Bayes classifier
1158:
1112:
1002:are independent given
979:
833:
798:
635:We seek to estimate P(
613:
584:
555:
514:
452:
262:
40:naive Bayes classifier
1261:
1227:at training time and
1222:
1159:
1092:
980:
834:
799:
627:is usually set at 1.
614:
585:
556:
515:
432:
242:
34:) is a probabilistic
1330:improve this article
1231:
1192:
1182:incremental learning
1020:
846:
811:
660:
594:
565:
527:
93:
1458:Bayesian estimation
46:The AODE classifier
1256:
1217:
1154:
1013:, it follows that
975:
829:
794:
609:
580:
551:
510:
390:
334:
207:
1406:
1405:
1398:
1380:
789:
539:
508:
462:
400:
335:
312:
272:
217:
152:
105:
16:(Redirected from
1470:
1437:
1427:Machine Learning
1419:
1401:
1394:
1390:
1387:
1381:
1379:
1338:
1314:
1306:
1265:
1263:
1262:
1257:
1252:
1251:
1226:
1224:
1223:
1218:
1213:
1212:
1163:
1161:
1160:
1155:
1147:
1146:
1128:
1127:
1111:
1106:
1088:
1087:
1060:
1059:
1044:
1043:
984:
982:
981:
976:
968:
967:
949:
948:
933:
932:
914:
913:
886:
885:
870:
869:
838:
836:
835:
830:
803:
801:
800:
795:
790:
788:
784:
783:
768:
767:
751:
747:
746:
731:
730:
708:
700:
699:
684:
683:
618:
616:
615:
610:
589:
587:
586:
581:
560:
558:
557:
552:
541:
540:
532:
519:
517:
516:
511:
509:
507:
503:
502:
490:
489:
477:
476:
464:
463:
455:
451:
446:
428:
427:
415:
414:
402:
401:
393:
389:
379:
378:
333:
326:
325:
310:
306:
305:
287:
286:
274:
273:
265:
261:
256:
238:
237:
219:
218:
210:
206:
196:
195:
150:
142:
141:
126:
125:
107:
106:
98:
21:
1478:
1477:
1473:
1472:
1471:
1469:
1468:
1467:
1443:
1442:
1441:
1440:
1429:, 58(1), 5–24.
1420:
1416:
1411:
1402:
1391:
1385:
1382:
1339:
1337:
1327:
1315:
1297:
1285:
1283:Implementations
1243:
1229:
1228:
1204:
1190:
1189:
1178:
1138:
1119:
1079:
1051:
1035:
1018:
1017:
1011:
1001:
994:
959:
940:
924:
905:
877:
861:
844:
843:
809:
808:
775:
759:
752:
738:
722:
709:
691:
675:
658:
657:
652:
645:
633:
592:
591:
563:
562:
525:
524:
494:
481:
468:
419:
406:
370:
317:
311:
297:
278:
229:
187:
151:
133:
117:
91:
90:
85:
78:
67:
60:
48:
23:
22:
15:
12:
11:
5:
1476:
1474:
1466:
1465:
1460:
1455:
1445:
1444:
1439:
1438:
1413:
1412:
1410:
1407:
1404:
1403:
1318:
1316:
1309:
1304:
1303:
1296:
1293:
1284:
1281:
1255:
1250:
1246:
1242:
1239:
1236:
1216:
1211:
1207:
1203:
1200:
1197:
1177:
1174:
1165:
1164:
1153:
1150:
1145:
1141:
1137:
1134:
1131:
1126:
1122:
1118:
1115:
1110:
1105:
1102:
1099:
1095:
1091:
1086:
1082:
1078:
1075:
1072:
1069:
1066:
1063:
1058:
1054:
1050:
1047:
1042:
1038:
1034:
1031:
1028:
1025:
1009:
999:
992:
986:
985:
974:
971:
966:
962:
958:
955:
952:
947:
943:
939:
936:
931:
927:
923:
920:
917:
912:
908:
904:
901:
898:
895:
892:
889:
884:
880:
876:
873:
868:
864:
860:
857:
854:
851:
828:
825:
822:
819:
816:
805:
804:
793:
787:
782:
778:
774:
771:
766:
762:
758:
755:
750:
745:
741:
737:
734:
729:
725:
721:
718:
715:
712:
706:
703:
698:
694:
690:
687:
682:
678:
674:
671:
668:
665:
650:
643:
632:
629:
608:
605:
602:
599:
579:
576:
573:
570:
550:
547:
544:
538:
535:
521:
520:
506:
501:
497:
493:
488:
484:
480:
475:
471:
467:
461:
458:
450:
445:
442:
439:
435:
431:
426:
422:
418:
413:
409:
405:
399:
396:
388:
385:
382:
377:
373:
369:
366:
363:
360:
357:
354:
351:
348:
345:
342:
338:
332:
329:
324:
320:
315:
309:
304:
300:
296:
293:
290:
285:
281:
277:
271:
268:
260:
255:
252:
249:
245:
241:
236:
232:
228:
225:
222:
216:
213:
205:
202:
199:
194:
190:
186:
183:
180:
177:
174:
171:
168:
165:
162:
159:
155:
148:
145:
140:
136:
132:
129:
124:
120:
116:
113:
110:
104:
101:
83:
76:
65:
58:
47:
44:
24:
14:
13:
10:
9:
6:
4:
3:
2:
1475:
1464:
1461:
1459:
1456:
1454:
1451:
1450:
1448:
1436:
1432:
1428:
1424:
1418:
1415:
1408:
1400:
1397:
1389:
1378:
1375:
1371:
1368:
1364:
1361:
1357:
1354:
1350:
1347: –
1346:
1342:
1341:Find sources:
1335:
1331:
1325:
1324:
1319:This article
1317:
1313:
1308:
1307:
1302:
1299:
1298:
1294:
1292:
1290:
1282:
1280:
1277:
1273:
1269:
1248:
1244:
1240:
1234:
1209:
1205:
1201:
1195:
1186:
1183:
1175:
1173:
1170:
1151:
1143:
1139:
1135:
1132:
1129:
1124:
1120:
1113:
1108:
1103:
1100:
1097:
1093:
1084:
1080:
1076:
1073:
1067:
1064:
1056:
1052:
1048:
1045:
1040:
1036:
1032:
1029:
1023:
1016:
1015:
1014:
1012:
1005:
998:
991:
972:
964:
960:
956:
953:
950:
945:
941:
937:
934:
929:
925:
918:
910:
906:
902:
899:
893:
890:
882:
878:
874:
871:
866:
862:
858:
855:
849:
842:
841:
840:
826:
823:
820:
817:
814:
791:
780:
776:
772:
769:
764:
760:
753:
743:
739:
735:
732:
727:
723:
719:
716:
710:
704:
696:
692:
688:
685:
680:
676:
672:
669:
663:
656:
655:
654:
649:
642:
638:
630:
628:
626:
622:
603:
597:
574:
568:
545:
533:
499:
495:
491:
482:
478:
473:
469:
456:
448:
443:
440:
437:
433:
424:
420:
416:
407:
394:
386:
383:
375:
371:
364:
361:
358:
355:
352:
349:
346:
343:
340:
336:
330:
327:
318:
313:
302:
298:
294:
291:
288:
283:
279:
266:
258:
253:
250:
247:
243:
234:
230:
226:
223:
211:
203:
200:
192:
188:
181:
178:
175:
172:
169:
166:
163:
160:
157:
153:
146:
138:
134:
130:
127:
122:
118:
114:
111:
99:
89:
88:
87:
82:
75:
71:
64:
57:
53:
45:
43:
41:
37:
33:
29:
19:
1426:
1417:
1392:
1383:
1373:
1366:
1359:
1352:
1340:
1328:Please help
1323:verification
1320:
1286:
1275:
1271:
1267:
1187:
1179:
1166:
1007:
1003:
996:
989:
987:
806:
647:
640:
636:
634:
624:
620:
522:
80:
73:
69:
62:
55:
51:
49:
31:
27:
26:
1447:Categories
1409:References
1386:March 2011
1356:newspapers
1287:The free
1130:∣
1094:∏
1049:…
951:∣
938:…
875:…
824:≤
818:≤
773:…
736:…
689:…
673:∣
604:⋅
575:⋅
546:⋅
537:^
487:′
479:∣
460:^
434:∏
412:′
398:^
384:≥
362:∧
356:≤
350:≤
337:∑
328:∈
323:′
314:∑
289:∣
270:^
244:∏
215:^
201:≥
179:∧
173:≤
167:≤
154:∑
131:…
115:∣
103:^
1295:See also
807:For any
1370:scholar
1372:
1365:
1358:
1351:
1343:
995:, ...
646:, ...
523:where
79:, ...
61:, ...
1377:JSTOR
1363:books
1349:news
1289:Weka
1006:and
68:, P(
32:AODE
18:AODE
1431:doi
1332:by
1449::
1425:.
839:,
639:|
590:,
72:|
1433::
1399:)
1393:(
1388:)
1384:(
1374:·
1367:·
1360:·
1353:·
1326:.
1276:k
1272:l
1268:n
1254:)
1249:2
1245:n
1241:k
1238:(
1235:O
1215:)
1210:2
1206:n
1202:l
1199:(
1196:O
1152:.
1149:)
1144:i
1140:x
1136:,
1133:y
1125:j
1121:x
1117:(
1114:P
1109:n
1104:1
1101:=
1098:j
1090:)
1085:i
1081:x
1077:,
1074:y
1071:(
1068:P
1065:=
1062:)
1057:n
1053:x
1046:,
1041:1
1037:x
1033:,
1030:y
1027:(
1024:P
1010:i
1008:x
1004:y
1000:n
997:x
993:1
990:x
973:.
970:)
965:i
961:x
957:,
954:y
946:n
942:x
935:,
930:1
926:x
922:(
919:P
916:)
911:i
907:x
903:,
900:y
897:(
894:P
891:=
888:)
883:n
879:x
872:,
867:1
863:x
859:,
856:y
853:(
850:P
827:n
821:i
815:1
792:.
786:)
781:n
777:x
770:,
765:1
761:x
757:(
754:P
749:)
744:n
740:x
733:,
728:1
724:x
720:,
717:y
714:(
711:P
705:=
702:)
697:n
693:x
686:,
681:1
677:x
670:y
667:(
664:P
651:n
648:x
644:1
641:x
637:y
625:m
621:m
607:)
601:(
598:F
578:)
572:(
569:P
549:)
543:(
534:P
505:)
500:i
496:x
492:,
483:y
474:j
470:x
466:(
457:P
449:n
444:1
441:=
438:j
430:)
425:i
421:x
417:,
408:y
404:(
395:P
387:m
381:)
376:i
372:x
368:(
365:F
359:n
353:i
347:1
344::
341:i
331:Y
319:y
308:)
303:i
299:x
295:,
292:y
284:j
280:x
276:(
267:P
259:n
254:1
251:=
248:j
240:)
235:i
231:x
227:,
224:y
221:(
212:P
204:m
198:)
193:i
189:x
185:(
182:F
176:n
170:i
164:1
161::
158:i
147:=
144:)
139:n
135:x
128:,
123:1
119:x
112:y
109:(
100:P
84:n
81:x
77:1
74:x
70:y
66:n
63:x
59:1
56:x
52:y
30:(
20:)
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
