Skip to main content
. 2010 Jul 20;10:37. doi: 10.1186/1472-6882-10-37

Table 2.

Results of syndrome models for inquiry diagnosis on total labels by using ML-kNN, RankSVM, BPMLL and kNN with different symptom subsets

symptoms Average_Precision(%)

ML-kNN kNN RankSVM BPMLL
125 76.2 ± 3.1 74.2 ± 3.3 70.9 ± 3.1 76.1 ± 3.8

106 76.6 ± 2.7 73.7 ± 3.3 71.0 ± 3.4 75.0 ± 3.3

83 76.8 ± 2.4 75.0 ± 3.1 74.3 ± 2.9 75.8 ± 3.4

64 76.6 ± 2.9 75.3 ± 2.9 74.4 ± 2.8 73.9 ± 3.9

52 78.0 ± 2.4 74.7 ± 2.3 73.3 ± 2.6 75.1 ± 2.7

32 75.7 ± 3.2 73.7 ± 3.5 72.1 ± 2.9 75.0 ± 2.7

21 74.9 ± 2.9 73.2 ± 3.8 70.5 ± 3.5 74.4 ± 3.3

symptoms Coverage

ML-kNN kNN RankSVM BPMLL

125 3.28 ± 0.32 3.44 ± 0.23 3.47 ± 0.28 3.30 ± 0.35

106 3.28 ± 0.27 3.41 ± 0.31 3.43 ± 0.28 3.52 ± 0.32

83 3.29 ± 0.28 3.46 ± 0.28 3.38 ± 0.29 3.32 ± 0.38

64 3.22 ± 0.23 3.43 ± 0.23 3.48 ± 0.29 3.41 ± 0.28

52 3.21 ± 0.24 3.43 ± 0.21 3.38 ± 0.35 3.34 ± 0.27

32 3.25 ± 0.31 3.49 ± 0.35 3.41 ± 0.25 3.43 ± 0.23

21 3.26 ± 0.32 3.51 ± 0.35 3.53 ± 0.36 3.42 ± 0.35

symptoms Ranking_Loss

ML-kNN kNN RankSVM BPMLL

125 0.290 ± 0.031 0.394 ± 0.044 0.384 ± 0.032 0.291 ± 0.036

106 0.283 ± 0.029 0.390 ± 0.037 0.351 ± 0.035 0.311 ± 0.031

83 0.277 ± 0.024 0.388 ± 0.037 0.329 ± 0.031 0.337 ± 0.029

64 0.266 ± 0.032 0.384 ± 0.042 0.348 ± 0.040 0.330 ± 0.027

52 0.271 ± 0.028 0.379 ± 0.034 0.353 ± 0.036 0.309 ± 0.048

32 0.273 ± 0.047 0.402 ± 0.036 0.343 ± 0.042 0.294 ± 0.029

21 0.279 ± 0.041 0.414 ± 0.029 0.369 ± 0.044 0.321 ± 0.037