I found the DoubleCrossValidation code of this paper, but I can't find the code where the final Selected features, Rank and N value are computed. I think this code computes only the ranks of inner and outer loops but the final chosen one is not clear or is not computed.
The code is here:
https://trac.nbic.nl/biomarkerfeatureselection/
https://trac.nbic.nl/biomarkerfeatureselection/downloads
The main problem is that I need the final rank and N value, to get top-N features from rank (from each method). Does anyone know where is the complete code or the complete selected features lists reported in this paper?
A critical assessment of feature selection methods for biomarker discovery in clinical proteomics.
Christin Christin, Huub C J Hoefsloot, Age K Smilde, Berend Hoekman, Frank Suits, Rainer Bischoff, Peter Horvatovich
P.S. All methods work for urine dataset but PCDA doesn't work for CSF dataset.
Article A Critical Assessment of Feature Selection Methods for Bioma...
The code is complete. Indeed the RP2cv_cc contains the code of the double cross-validation version of the PCADA, and the single-crossvalidation code determining the final selected features is located in the function analyse_pcda_result. Run first PCDA with double cross-validation (RP2cv_cc) and use the results in analyse_pcda_result and you should get the list of selected features in the resulting structure.
- Badges
- Science topic
- Similar topics
- Engineering
- Control Theory
- Control Systems
- Control Systems Engineering