I have a classification task. I want to use apache spark ml lib SVM algorithm for classification. I have input data which is n-dimensional. In the feature vectors some of dimensions may be missing.
How to approach with missing values? I think it would me wrong to assume missing values as zero or something else.