BMC Medical Informatics and Decision Making

Table 2 Results of a 10-fold cross-validation run for baseline and HMM classifiers

From: Mining Adverse Drug Reactions from online healthcare forums using Hidden Markov Model

					Baseline classifier
Iteration	Train set	Test set	True	False	True	False	Precision	Recall	F-Score	Accuracy
			positive	positive	negative	negative
1	1761	196	27	8	135	26	0.771	0.509	0.614	0.827
2	1761	196	25	15	137	19	0.625	0.568	0.595	0.827
3	1761	196	35	14	125	22	0.714	0.614	0.660	0.816
4	1761	196	27	14	130	25	0.659	0.519	0.581	0.801
5	1761	196	28	9	130	29	0.757	0.491	0.596	0.806
6	1761	196	27	21	131	17	0.563	0.614	0.587	0.806
7	1761	196	23	17	137	19	0.575	0.548	0.561	0.816
8	1761	196	27	17	128	24	0.614	0.529	0.568	0.791
9	1761	196	18	12	140	26	0.6	0.409	0.486	0.806
10	1763	194	28	10	128	28	0.737	0.5	0.596	0.804
					HMM classifier
Iteration	Train set	Test set	True	False	True	False	Precision	Recall	F-Score	Accuracy
			positive	positive	negative	negative
1	1761	196	42	8	136	10	0.84	0.808	0.824	0.908
2	1761	196	30	10	142	14	0.75	0.682	0.714	0.878
3	1761	196	50	10	123	13	0.833	0.794	0.813	0.883
4	1761	196	29	5	142	20	0.853	0.592	0.699	0.872
5	1761	196	37	14	135	10	0.725	0.787	0.755	0.878
6	1761	196	40	9	140	7	0.816	0.851	0.833	0.918
7	1761	196	37	11	131	17	0.771	0.685	0.725	0.857
8	1761	196	44	8	131	13	0.846	0.772	0.807	0.893
9	1761	196	29	16	143	8	0.644	0.784	0.707	0.878
10	1763	194	34	9	135	16	0.791	0.68	0.731	0.871

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com