Skip to main content
Figure 3 | BMC Bioinformatics

Figure 3

From: Unsupervised reduction of random noise in complex data by a row-specific, sorted principal component-guided method

Figure 3

Number of top PCs to keep in PCA for noise reduction. The simulated data set used in Figure 2 was subjected to noise reduction by PCA. (A) The relative variance associated with each PC. The total variance of the data set was scaled to 100%. The bars for PC3 and PC9 are labeled because the variance values with the PCs immediately after them dropped noticeably. (B) The noise RMS ratio resulting from PCA with different numbers of top PCs kept. The values for two data points at PCs1-12 and PCs1-26 are indicated in the plot. The noise RMS ratio value of 0.76 at PCs1-26 was the best when PCA was used. The dashed horizontal line at the noise RMS ratio value of 0.63 indicates the noise RMS ratio value that was achieved by RSPR-NR.

Back to article page