I want to compare the classification performance of Random forest with variable selection algorithm (method A) and Random forest only (method B). The original dataset has 100 variables. After using variable selection, method A used only 27 variables and achieve 89% overall accuracy while B is 87.2% overall accuracy using all original variables.
I used the McNemar's chi-square test and z value from McNemar's chi-square test z=0.7. Therefore, Method A and B is no significant difference. Is the statement that the variable selection algorithm can improve the classification accuracy with fewer variable not correct?
Though the performance of classifiers depends upon noise free features but what feature selection scheme is suitable for the said dataset requires a comprehensive experimentation. So, i suggest to use various features selection schemes with different number of features to identify useful features.
Please find these comparison using machine learning .
Brief Bioinform. 2019 Mar; 20(2): 492–503.Published online 2017 Oct 16. doi: 10.1093/bib/bbx124PMCID: PMC6433899PMID: 29045534
Evaluation of variable selection methods for random forests and omics data sets
Frauke Degenhardt,1 Stephan Seifert,2 and Silke Szymczak2Author information Article notes Copyright and License information Disclaimer
This is the link below .
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6433899/
- Badges
- Science topic
- Similar topics
- Correction