It is hard to identify interaction between two variables by looking at the variable importance plot. One way is to use the partial dependence plot. I was wondering if anyone has experience by using two variables at a time to detect interaction between them for the corresponding dependent variable.
Thanks!
The following link would provide a good answer to your question:
http://stats.stackexchange.com/questions/21152/obtaining-knowledge-from-a-random-forest
I hope this satisfies you.
I also suggest taking a look at Ishwaran et al. 2010.
http://www.ccs.miami.edu/~hishwaran/papers/IKGML.JASA.2010.pdf
The methods therein can be found in the 'randomForestSRC' package. It should be noted that a recent paper indicated that identifying interactions, at least in the hyper-dimensional space of gene interactions, can be pretty tough:
https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-016-0995-8
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package