Skip to main content

Table 2 This table shows the main results comparing the feature selection benchmarks of our approach with Lasso and 1-SVM on the spiked data-set. Given results correspond to the average results over 10 repetitions of the classifier with 6 non-zero values

From: Sparse Proteomics Analysis – a compressed sensing-based approach for feature selection and classification of high-dimensional proteomics mass spectrometry data

  SPA Lasso 1-SVM
Concentration TP [a] Sens [b] Specs [c] B. Acc [d] TP Sens Spec B. Acc TP Sens Spec B. Acc
0.075pMol/L 2 0.333 1.000 0.667 1 0.167 1.000 0.583 1 0.167 1.000 0.583
3.03pMol/L 4 0.667 1.000 0.833 2 0.333 1.000 0.667 1 0.167 1.000 0.583
0.30nMol/L 2 0.333 1.000 0.667 1 0.167 1.000 0.583 1 0.167 1.000 0.583
0.76nMol/L 2 0.333 1.000 0.667 2 0.333 1.000 0.667 2 0.333 1.000 0.667
121.21nMol/L 3 0.500 1.000 0.750 2 0.333 1.000 0.667 2 0.333 1.000 0.667
  1. [a]TP: Number of spiked peaks that are correctly detected
  2. [b]Sens: Sensitivity in detecting spiked peaks (T P/(T P+F N))
  3. [c]Spec: Specificity in detecting spiked peaks (T N/(F P+T N))
  4. [d]B. Acc: Balanced Accuracy (\( \frac {\text {sens.} + \text {spec.}}{2}\))