Imputation of Missing Clinical Covariates for Downstream Classification Problems

Noticeable growth in the use of intelligent devices has resulted generation vast amounts data from sensor devices. When dealing with large data, it is common to observe databases missing values. This a challenge for miners because various methods analysis only work well on complete databases. A traditional approach handling discard instances values and cases analysis. However, research shown that this not practical especially when are missing. led an increased need develop strategies replacing plausible through imputation. study presents imputation strategy called med.BFMVI recovering before training downstream classification models. Experiments simulated missingness 10% 40% using MCAR MAR mechanisms performance proposed technique was measured against state-of-the-art techniques. Overall, algorithm recorded best accuracy as opposed benchmark techniques showed significant improvements learning.