决策代价约简求解中的交叉验证策略

张龙波,李智远,杨习贝,王怡博

Cross⁃validation strategy in attribute reduction based on decision cost

Longbo Zhang,Zhiyuan Li,Xibei Yang,Yibo Wang

表3 分类精度和约简长度对比

Table 3 Classification accuracies and lengths of feature subsets

ID KNN SVM CART 约简长度 算法1 算法2 算法1 算法2 算法1 算法2 算法1 算法2 1 0.6087 0.6043 0.5315 0.5387 0.6147 0.6318 2.14 2.58 2 0.8652 0.9196 0.5290 0.5384 0.7230 0.7565 5.78 9.82 3 0.8944 0.9185 0.9002 0.9270 0.8956 0.9300 5.70 7.44 4 0.8154 0.8647 0.8068 0.8381 0.7721 0.8078 4.28 5.64 5 0.8152 0.8458 0.8223 0.8380 0.7816 0.8003 2.82 4.64 6 0.8958 0.8975 0.8581 0.8618 0.8538 0.8806 2.32 3.80 7 0.6351 0.6672 0.5316 0.5507 0.6117 0.6477 3.42 5.04 8 0.7874 0.8163 0.7453 0.7914 0.7671 0.7982 3.30 4.78