A multi-manifold learning based instance weighting and under-sampling for imbalanced data classification problems

Journal of Big Data

Table 6 The performance of the 3NN classifier with the multi-manifold approach with step of 5 percent

Dataset	Recall	Precision	G-means	F-measure	Accuracy
ecoli1	0.93 ± 0.130	0.70 ± 0.201	0.89 ± 0.110	0.85 ± 0.153	0.87 ± 0.111
ecoli2	0.95 ± 0.111	0.78 ± 0.188	0.94 ± 0.074	0.91 ± 0.105	0.94 ± 0.063
ecoli3	0.93 ± 0.225	0.60 ± 0.337	0.90 ± 0.152	0.75 ± 0.231	0.90 ± 0.097
ecoli4	0.95 ± 0.150	0.75 ± 0.334	0.93 ± 0.178	0.88 ± 0.279	0.91 ± 0.199
ecoli0147vs56	0.88 ± 0.183	0.72 ± 0.251	0.92 ± 0.106	0.84 ± 0.159	0.96 ± 0.036
ecoli034_5	0.90 ± 0.200	0.78 ± 0.269	0.92 ± 0.122	0.86 ± 0.180	0.95 ± 0.067
ecoli0147_2356	0.82 ± 0.240	0.66 ± 0.253	0.87 ± 0.149	0.81 ± 0.159	0.94 ± 0.047
glass0	0.89 ± 0.124	0.64 ± 0.116	0.79 ± 0.070	0.78 ± 0.076	0.78 ± 0.083
glass0123456	0.96 ± 0.80	0.86 ± 0.227	0.94 ± 0.117	0.92 ± 0.153	0.93 ± 0.133
kddcup-buffer_overflow_vs_back	1 ± 0	1 ± 0	1 ± 0	1 ± 0	1 ± 0
new-thyroid1	0.93 ± 0.200	0.93 ± 0.155	0.95 ± 0.125	0.93 ± 0.160	0.97 ± 0.048
page-blocks-1-3_vs_4	0.93 ± 0.133	0.75 ± 0.293	0.94 ± 0.079	0.87 ± 0.210	0.95 ± 0.069
Pima	0.62 ± 0.208	0.56 ± 0.148	0.64 ± 0.114	0.58 ± 0.141	0.68 ± 0.103
segment0	0.99 ± 0.027	0.92 ± 0.056	0.99 ± 0.014	0.97 ± 0.023	0.99 ± 0.010
shuttle_2_vs_5	1 ± 0	0.91 ± 0.274	0.99 ± 0.025	0.97 ± 0.100	0.98 ± 0.047
vehicle2–1	0.97 ± 0.042	0.82 ± 0.098	0.94 ± 0.030	0.92 ± 0.051	0.93 ± 0.036
vowel0	0.90 ± 0.213	0.67 ± 0.330	0.89 ± 0.181	0.87 ± 0.157	0.89 ± 0.153
Wisconsin	0.98 ± 0.027	0.93 ± 0.083	0.94 ± 0.074	0.97 ± 0.050	0.94 ± 0.061
Average	0.918	0.767	0.91	0.871	0.917