Hi,
I have experience using R for predictive modeling. I would first analyze the different variables and find their predictive power, drop any features that do not help in prediction, then try different ml algos. then build the final model and evaluate it.
Regards,
Chris