I've very new to SVMs so excuse me if this question is misinformed.
Say I wanted to predict the outcome of an event such as "will a given golfer land a drive on the green on his first swing" say for this golfer I have a great deal of historical data as well as a huge number of data points including everything from wind speed to the golfers sock color at the time of the swing.
If I train the SVM with the historical data is there a way to determine what the best weightings for the vector components should be. For example wind speed maybe a large contributor to the prediction where as sock color may have no effect on the prediction .
From what I've been reading the vector components being fed in the SVM have to be weighted ahead of time or they are all given a weighting of 1. Is this true? Is weighting the vector components even required? If they are what is the proper way to determine the weights?
[link][2 comments]