Improved KNN Imputation for Missing Values in Gene Expression Data

The problem of missing values has long been studied by researchers working in areas data science and bioinformatics, especially the analysis gene expression that facilitates an early detection cancer. Many attempts show improvements made excluding samples with information from process, while others have tried to fill gaps possible values. While former is simple, latter safeguards loss. For that, a neighbour-based (KNN) approach proven more effective than other global estimators. paper extends this further introducing new summarization method KNN model. It first study applies concept ordered weighted averaging (OWA) operator such context. In particular, two variations OWA aggregation are proposed evaluated against their baseline neighbor-based models. Using different ratios 1%–20% set six published datasets, experimental results suggest methods usually provide accurate estimates those compared methods. Specific rates 5% 20%, best NRMSE scores as averages across datasets 0.65 0.69, highest measures obtained existing techniques included 0.80 0.84, respectively.