Fig. 1From: Unsupervised software defect prediction using median absolute deviation threshold based spectral classifier on signed Laplacian matrixEigenvectors distribution. The eigenvectors distributions in a are obtained from signed Laplacian matrix decomposition of all datasets using Eq. 10. These eigenvectors distributions show that almost eigenvectors come with outliers. b shows the example of eigenvector distribution with MAD of CM1 datasetBack to article page