Other than regression (some of it quite advanced) my attempts to use ML have been a complete failure! The few times I have found something informative I always find that there is some form of look-forward bias that invalidates the results.
The regression I do in R is really nice though. Extremely fast and not something I could ever do in excel. As is often the case, getting the data ready is far more work than running the analysis. R's 'tidymodels' makes it easy to run dozens of different models against dozens of different datasets, in one block of code.