Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: Missing value imputation improves clustering and interpretation of gene expression microarray data

Figure 1

A schematic illustration of the comparison procedure. The testing procedure was repeated for each of the eight datasets (see Table 1) and for each of the missing value rates (q = 0.5%, 1%, 5%, 10%, 15%, and 20%). The different imputation methods (see Table 2) were evaluated in terms of their capability to reproduce the original data values (NRMSE), clustering solutions (ADBP for gene clusters) and their biological interpretations (ADBP for GO terms). Clustering analyses were performed also directly on datasets with missing values (Nimp). Missing values were generated 30 times and the k-means clustering was performed for multiple numbers of clusters (k = 2, 3,..., 10).

Back to article page