Adding progress bar or percentage to tune function in R

1
votes

In e1071 library of R has an inbuilt tune() function to optimize our algorithm.

tuneop=tune(svm,y~.,data=dat,kernel="radial",ranges=list(cost=c(0.001, 0.01, 0.1, 1,5,10,100),gamma=c(0.001, 0.01, 0.1, 1,5,10)))

I use tune function to optimize my gamma and cost parameter in my SVM model. For small dataset tune() requires only a small amount of time to generate bestmodel . But for a very large data set it takes a lot of time. So is it possible to add a progress bar or percentage to monitor the progress of the tuning of our model.

2
rmachine-learningprogress-barsvm
Progress bars usually used with a loops. – Artem Klevtsov
For very large sets SVMs may not be applicable at all. The computational costs scale approximately with N^3 if N is the number of observations. – RHertel

2 Answers

2
votes

Not exactly what you're looking for, but I use the beep() function from the beepr package. It makes a sound when a command is finished.

2
votes

As far as I'm aware e1071 doesn't include a option for a progress bar or verbose mode in it's tuning method.

However the tune method is a simple grid search, from the documentation

This generic function tunes hyperparameters of statistical methods using a grid search over supplied parameter ranges.

So it would be fairly simple for you to write your own tuning function.

Otherwise you could use a package which wraps e1071 and implements it's own tuning method with better verbosity, such as mlr or caret.