Doing Machine Learning In R

less than 1 minute read

Doing machine learning in R

I recently spoke to the EdinbR R user group about predictive modelling and machine learning in R. The slides are attached available to download here. The talk was a very high-level overview of how to actually train predictive models in R, and I’d probably summarise the following as the key points:

  • You can fit most commonly used machine learning tools in R: you don’t have to learn Python
  • Many individual machine learning algorithms in R exist in separate packages, with varying syntax and input / output, making training painful for multiple models
  • Meta-packages (CRAN terminology) like caret and mlr provide high-level interfaces to a huge variety of ML algorithms
  • caret streamlines the process of cross-validation, prediction and calculating accuracy