I dont understand when is it appropriate to standardize my data and when it is appropriate to normalize my data. I know it will differ by project of course, but for my case, I do not know what type of things or 'rule of thumb' to consider. Right now, I"m working on a spreadsheet with 12 numerical variables. All > 0, and some range from 0 to 5, some 1 to 15, some 1 to 50. My objective is to pre-process my data so I can get a good validation set to train and do a prediction on my test set. While all my variables are continuous, my response variable is categorical, either 1, 2, or 3.

Where would I start?