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 8 不同缺失率下不同插补方法在AD数据集中的性能对比结果
Tab.8 Performance comparison results of different interpolation methods under different missing rates in AD dataset
方法${p_{{\mathrm{mis}}}}$=10%${p_{{\mathrm{mis}}}}$=30%${p_{{\mathrm{mis}}}}$=50%
RMSERERMSERERMSERE
平均值0.132±0.0060.629±0.0130.131±0.0020.629±0.0080.131±0.0030.627±0.006
ICE0.124±0.0060.577±0.0090.123±0.0010.578±0.0090.126±0.0030.581±0.008
EM0.124±0.0060.567±0.0070.123±0.0010.572±0.0070.126±0.0030.578±0.007
GAIN0.135±0.0060.629±0.0070.137±0.0020.650±0.0040.205±0.0320.668±0.006
MissForest0.118±0.0060.560±0.0100.119±0.0020.566±0.0060.123±0.0030.573±0.006
MIWAE0.137±0.0070.500±0.0100.137±0.0020.499±0.0070.137±0.0040.499±0.006
TabCSDI0.104±0.0080.388±0.0100.109±0.0030.403±0.0080.115±0.0040.419±0.005
PNDM_Tab0.110±0.0070.384±0.0090.113±0.0030.401±0.0050.119±0.0040.423±0.005