基于加速扩散模型的缺失值插补算法
王圣举,张赞

Missing value imputation algorithm based on accelerated diffusion model
Shengju WANG,Zan ZHANG
表 4 不同插补方法在混合特征数据集中的性能对比结果
Tab.4 Performance comparison results of different imputation methods in mixed feature datasets
方法ADHeartStudents
RMSERERMSERERMSERE
平均值0.131±0.0030.628±0.0100.164±0.0060.487±0.0210.231±0.0070.531±0.014
ICE0.122±0.0030.571±0.0110.145±0.0050.391±0.0240.186±0.0110.432±0.013
EM0.122±0.0030.564±0.0060.145±0.0060.393±0.0100.187±0.0100.414±0.010
GAIN0.135±0.0020.637±0.0070.158±0.0030.403±0.0190.257±0.0020.488±0.014
MissForest0.118±0.0030.560±0.0050.140±0.0040.336±0.0260.169±0.0060.414±0.013
MIWAE0.136±0.0040.500±0.0030.185±0.0130.477±0.0400.305±0.0090.528±0.007
TabCSDI0.107±0.0040.393±0.0060.147±0.0020.389±0.0320.221±0.0130.402±0.010
PNDM_Tab0.111±0.0030.391±0.0020.139±0.0040.351±0.0270.190±0.0090.343±0.012