SVM machine learning

Question

What does it mean by "Features of datasets should vary on same scale in SVM Machine Learning Models for getting better accuracy on test set data"

Answer 1

This excerpt lacks the context a bit. Most probably it just means that scale of features should not be extremely different. For example, imagine you have feature 1 with range of values between [0, 10^5] and feature 2 spanning [0-1]. You need to normalize them to make in the same scale. zscoring would be one good option. This is true also in multivariate methods like principal component analysis.