Max Lin on finishing second in the R Challenge

I participated in the R package recommendation engine competition on Kaggle for two reasons. First, I use R a lot. I cannot learn statistics without R. This competition is my chance to give back to the community a R package recommendation engine. Second, during my day job as an engineer behind a machine learning service in the cloud, product recommendation is one of the most popular applications our early adopters want to use the web service for. This competition is my opportunity to stand in users’ shoes and identify their pain points associated with building a recommendation system. I treat R package recommendation as a binary classification task: a classifier $latex f(u, p)$ takes as input a user $latex u$ and a package $latex p$, and predicts whether the user will install the package o