# LIBSVM "Warning: using -h 0 may be faster"

I am using LIBSVM package (Matlab version) in order to perform wind power forecasting. My training dataset is quite big, approx. 52000 values of 4 parameters.

The code seems to run correctly but is quite slow and I get the following message:

WARNING: using -h 0 may be faster


What does that mean and what do I have to change in my code?

• Add the -h 0 flag to your options string. Basically it means that no shrinking heuristics will be used by the solver. Sep 17 '13 at 11:38

This means, that optimization algorithm detected that with high probability (not in the strict, mathematical sense) you can speed up your training by turning the -h 0 flag in your options. Basically, -h is the shrinking heuristics, implemented in the libsvm package which for some data significantly reduces number of required computations, while in others - makes it slower. There is no general rule, but there are some heuristics regarding this heuristic, which "detect" when it could help - and this is exactly the reason for this message - one of such "metaheuristics" detected, that for your particular data and parameters - it could be more valuable to set the -h 0 flag.
• For folks (like me) reading this and using sklearn, use shrinking=False.