We recently came across the article, "Random Forest---the go-to machine learning algorithm" from TechWorld Australia.Read More
Leo Breiman laid out his philosophy of data analysis and method invention in his 2001 paper in Statistical Science, "Statistical Modeling: The Two Cultures" in which, to over simplify, he says: if a methodology can generate high predictive accuracy on test data sets we do not need to provide any further justification or defense or explanation. Predictive accuracy is sufficient to justify whatever we do to reach a predictive model. In particular, we do not need a theory describing the process that generates the data we are studying and we do not a need theory explaining the learning machine we are using. He certainly took the latter position regarding one his greatest achievements: Random Forests (co-developed with Adele Cutler).