Data Science Stack Exchange is a question and answer site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field. It only takes a minute to sign up.
Sign up to join this community
I'm going through Wikipedia article on SVM and came across these equations:
It says:
With a normalized or standardized dataset, the parallel hyperplanes to our required hyperplane $wx+b=0$, can be described by the equations $wx+b=1$ and $wx+b=-1$
What is meant by normalized and standardized data here? If the data is not standardized, what equation should be used?
Normalized / standardized data are necessary to train an SVM classifier. Before running an SVM you should always scale your variables: it can be done either through min-max scaling or with standardization (in statistics are called Z-scores).
It's a fundamental dataprep operation, without it a classifier such as SVM wouldn't be useful. That is because if variables are not on the same scale the ones with the larger variance will dominate all the others in the training.
The equations that you quoted are the support vectors. In the image next to them: the dashed lines whose distance must be maximixed by the SVM classifier. The values of 1 and -1 are the labels that the model is attributing to the target classes (in a binary classification problem).
Hope this helps, otherwise let me know.
