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 | AD | | Heart | | Students | RMSE | RE | | RMSE | RE | | RMSE | RE | 平均值 | 0.131±0.003 | 0.628±0.010 | | 0.164±0.006 | 0.487±0.021 | | 0.231±0.007 | 0.531±0.014 | ICE | 0.122±0.003 | 0.571±0.011 | | 0.145±0.005 | 0.391±0.024 | | 0.186±0.011 | 0.432±0.013 | EM | 0.122±0.003 | 0.564±0.006 | | 0.145±0.006 | 0.393±0.010 | | 0.187±0.010 | 0.414±0.010 | GAIN | 0.135±0.002 | 0.637±0.007 | | 0.158±0.003 | 0.403±0.019 | | 0.257±0.002 | 0.488±0.014 | MissForest | 0.118±0.003 | 0.560±0.005 | | 0.140±0.004 | 0.336±0.026 | | 0.169±0.006 | 0.414±0.013 | MIWAE | 0.136±0.004 | 0.500±0.003 | | 0.185±0.013 | 0.477±0.040 | | 0.305±0.009 | 0.528±0.007 | TabCSDI | 0.107±0.004 | 0.393±0.006 | | 0.147±0.002 | 0.389±0.032 | | 0.221±0.013 | 0.402±0.010 | PNDM_Tab | 0.111±0.003 | 0.391±0.002 | | 0.139±0.004 | 0.351±0.027 | | 0.190±0.009 | 0.343±0.012 |
|
|
|