Tuning parameters in R isn't a new topic.

I am embarking on some modelling of a large data set, with the goal to create a predictive model using R, selecting predictors analytically from an abundance: ~1000 observations of ~200 predictors. I will be looking at various methods, for example boosting.

All the methods I am looking at require fine-tuning of model parameters and I would like to optimise this process as far as possible - both in terms of effectiveness and computational cost. In this video on gradient boosting - which is all in Python - I saw a function in the scikit-learn module in python called GridSearchCV that accepts a range for all model parameters and works through all of them to find the optimal set of parameters specific to your model, based on how you define optimal.

Is the established way of achieving this in R to use the caret package and train methodology? This seems very comparable to the Python method linked above, but are there different ways? Literature references?

  • $\begingroup$ let tell us your model and the assumptions. There are hopefully fast and reliable way of optimizing tuning parameter. $\endgroup$ – TPArrow Nov 27 '15 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.