[R] Please point me in the right direction: decision trees or possibly something better

The Linear Baseline model solves a convex problem and thus will converge to roughly the same optimum. This gives you basically an Idea for how informative naive correlation between features are. Logistic Regressions and Support Vector Machines are a common choice here in my experience. Regarding nonlinear models, the random forrest classifier is neat. XGBoost (Gradient Boosted Decision forrest) from the package of the same name is also realy good.