So, out of existing classification algorithms such as Support Vector Machine (SVM), K-nearest neighbor (KNN), Interval Valued Classification (IVC) and the improvised Interval Value based Particle Swarm Optimization (IVPSO) algorithm which one is potentially robust in handling, reliable in terms of results (not necessarily popular!) and why ?
I suggest you a paper MAQC-II Nature biotechnology (http://www.ncbi.nlm.nih.gov/pubmed/20676074) !. The methodology is to implement all of them and test wich one is the best. You can use a combination of two. The final decision depends on the crossvalidation results in your data.
From what I have seen I would say there is no overall best classification algorithm. It really depends on the data in hand. I would even say it is more important to understand the data and the underlying situation than choosing the classification algorithm. I also agree with Patricia, you should always try several classification methods in combination with several preprocessing (normalization) and feature selection methods and always (!) use resampling methods (cross-validation or bootstrap) to compare and validate the results. There is a lot of interesting material especially reprints and presentations under http://linus.nci.nih.gov/
Best
Matthias
We wrote a paper comparing several methods sometime ago. The paper was discussing methods to remove batch effect across independent studies. The machine learning techniques discussed are very similar.
On another subject I just published a video to explain our website. I would really appreciate your feedback and please share it if you find it relevant.
Best,
Alain
Article GENESHIFT: a Non-Parametric Approach for Integrating Microar...
- Badges
- Science topic
- Similar topics
- Clinical Pharmacology
- Dosing